Skip to contentSkip to conversation

Disgusted with kbuild developers

DISCUSSION Jeff Garzik 37 participants 137 messages Feb 15, 2002 – Feb 20, 2002

lore
Share & tools
lore
#0 lore raw

Ladies and gentlemen,

I would find this pathetic, if it didn't make me so mad. Making an end run around the system, are we, Eric?

Small wonder the message below didn't make it to linux-kernel:
suggestions would have been made that CML2 is DOA.

ESR's message to the kbuild list:

	http://marc.theaimsgroup.com/?l=kbuild-devel&m=101373619625183&w=2
The rest of the thread:
	http://marc.theaimsgroup.com/?t=101373623000001&r=1&w=2

For an open source "guru", Eric, you sure seem to like to turn to cronyism and secret meetings when the going gets tough.

Jeff

Signature · 3 lines
Jeff Garzik      | "I went through my candy like hot oatmeal
Building 1024    |  through an internally-buttered weasel."
MandrakeSoft     |             - goats.com
#1 lore raw
Quoted 7 lines · Jeff Garzik

ESR's message to the kbuild list:

	http://marc.theaimsgroup.com/?l=kbuild-devel&m=101373619625183&w=2
The rest of the thread:
	http://marc.theaimsgroup.com/?t=101373623000001&r=1&w=2

For an open source "guru", Eric, you sure seem to like to turn to cronyism and secret meetings when the going gets tough.

This makes me wonder whether Eric works in a cathedral or in an ivory tower ...

Rik

Signature · 4 lines
"Linux holds advantages over the single-vendor commercial OS"
    -- Microsoft's "Competing with Linux" document

http://www.surriel.com/		http://distro.conectiva.com/
#3 lore raw
In reply to Jeff Garzik

Jeff Garzik <[email protected]>:

Quoted 2 lines

I would find this pathetic, if it didn't make me so mad. Making an end run around the system, are we, Eric?

No. I'm doing what Dirk Hohndel recommended I do during a phone call that happened at his initiative, last Friday morning.

What "system" would you be referring to, anyway, Jeff? Is there some reason a respected open-source developer like Dirk who has concerns should not have a conversation with Linus to address problems he thinks are significant? Is there some reason I should not have asked the kbuild team members to give him appropriate background?

I don't understand what is upsetting you. Is there some rule that all conversations with Linus must go through Jeff Garzik? If so, I was never informed of it.

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#4 lore raw
In reply to Jeff Garzik
Quoted 3 lines · Eric S. Raymond

I don't understand what is upsetting you. Is there some rule that all conversations with Linus must go through Jeff Garzik? If so, I was never informed of it.

No, but at the least keeping Linux-Kernel in the loop would be considered not just nice, but a somewhat more courteous method than 'sending someone around to go see Linus' when your pet project isn't getting the acceptance you hoped for.

Signature · 2 lines
| Dave Jones.        http://www.codemonkey.org.uk
| SuSE Labs
#5 lore raw
In reply to Jeff Garzik
Quoted 9 lines · Eric S. Raymond

Jeff Garzik <[email protected]>:

Quoted 2 lines

I would find this pathetic, if it didn't make me so mad. Making an end run around the system, are we, Eric?

What "system" would you be referring to, anyway, Jeff? Is there some reason a respected open-source developer like Dirk who has concerns should not have a conversation with Linus to address problems he thinks are significant? Is there some reason I should not have asked the kbuild team members to give him appropriate background?

Yeah, there is. The point of *open* source is that it is *open*, Eric. Jeff, if I understand him correctly, is making the point that if a system can't stand up to public scrutiny, trying to get it in by private campaigning is lame, not the open source way, and a little bit sleazy.

The open source development model depends on peer review, you might go back and read some of your many essays on the topic. Don't you take commercial companies to task for doing exactly what you are doing? Your actions are confusing, do your writings apply to other people but not to you? If there is some misunderstanding about your actions, please clear it up. If not, practice what you preach.

Signature · 2 lines
---
Larry McVoy            	 lm at bitmover.com           http://www.bitmover.com/lm 
#6 lore raw
In reply to Jeff Garzik
Quoted 4 lines · Eric S. Raymond

What "system" would you be referring to, anyway, Jeff? Is there some reason a respected open-source developer like Dirk who has concerns should not have a conversation with Linus to address problems he thinks are significant?

Is there some reason this isn't being discussed on l-k again? Lots of people have concerns about the whole build system. Ignoring the whole python debate, there's actual issues about the CML2 package.

Signature · 2 lines
Tom Rini (TR1265)
http://gate.crashing.org/~trini/
#7 lore raw
In reply to Jeff Garzik
Quoted 14 lines · Eric S. Raymond

Jeff Garzik <[email protected]>:

Quoted 2 lines

I would find this pathetic, if it didn't make me so mad. Making an end run around the system, are we, Eric?

No. I'm doing what Dirk Hohndel recommended I do during a phone call that happened at his initiative, last Friday morning.

What "system" would you be referring to, anyway, Jeff? Is there some reason a respected open-source developer like Dirk who has concerns should not have a conversation with Linus to address problems he thinks are significant? Is there some reason I should not have asked the kbuild team members to give him appropriate background?

Quoted 3 lines

I don't understand what is upsetting you. Is there some rule that all conversations with Linus must go through Jeff Garzik? If so, I was never informed of it.

Eric,

What I see on linux-kernel is you writing long diatribes, and then promptly ignoring feedback from kernel developers who would actually be using your system. After months and months of being told that Configure.help needed to be split up, you just continued to whine. Linus finally sat down and did it himself. Back in December, Linus talked about per-driver and/or per-subsystem metadata files, which would contain (among other things) the config information. That seems like a sane option to me, and other developers agreed. You ignored that too.

With regards to kbuild (not CML2), Keith publicly stated he didn't care about 2.5.

I feel sorry about that you have troubled Dirk, without giving him all the facts.

Jeff

Signature · 3 lines
Jeff Garzik      | "I went through my candy like hot oatmeal
Building 1024    |  through an internally-buttered weasel."
MandrakeSoft     |             - goats.com
#2 lore raw
In reply to Rik van Riel
Quoted 7 lines · Rik van Riel
Quoted 5 lines · Jeff Garzik

ESR's message to the kbuild list:

	http://marc.theaimsgroup.com/?l=kbuild-devel&m=101373619625183&w=2
The rest of the thread:
	http://marc.theaimsgroup.com/?t=101373623000001&r=1&w=2
Quoted 5 lines
Quoted 2 lines

For an open source "guru", Eric, you sure seem to like to turn to cronyism and secret meetings when the going gets tough.

This makes me wonder whether Eric works in a cathedral or in an ivory tower ...

Judging by usual Eric's intellectual erm... output the word you are looking for is "outhouse".

#8 lore raw
In reply to Jeff Garzik

Larry McVoy <[email protected]>:

Quoted 4 lines

Yeah, there is. The point of *open* source is that it is *open*, Eric. Jeff, if I understand him correctly, is making the point that if a system can't stand up to public scrutiny, trying to get it in by private campaigning is lame, not the open source way, and a little bit sleazy.

So what "private campaigning" is going on here, exactly?

Dirk approached me because he was concerned that politics and poor communication might be getting in the way of some valuable work. He asked me to have the kbuild team fill him in on the background, and I relayed that request. The kbuild team's discussion with Dirk is taking place on a public mailing list.

I'm bewildered. What's the problem here?

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#9 lore raw
In reply to Jeff Garzik

Jeff Garzik <[email protected]>:

Quoted 4 lines

What I see on linux-kernel is you writing long diatribes, and then promptly ignoring feedback from kernel developers who would actually be using your system. After months and months of being told that Configure.help needed to be split up, you just continued to whine.

I heard Linus say that he wanted Configure.help split up. I agreed to do it, too, in CML2. But I thought I was not supposed to mess with existing formats -- in fact, I was afraid to, because you and Al Viro beat me up so hard about changing *nothing* before cutover. The CML1 code owns that format, and I am not the CML1 maintainer. In fact mec specifically turned me down when I volunteered for that job; he said he wanted me concentrating on CML2.

Neither Linus nor anyone else ever said to me "Eric, it's your job to make that change." When I asked for guidance, Linus blackholed my mail. Was I supposed to be practicing telepathy, here?

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#135 lore raw
In reply to Jeff Garzik
Quoted 2 lines · Jeff Garzik

For an open source "guru", Eric, you sure seem to like to turn to cronyism and secret meetings when the going gets tough.

Yeah, you should have heard him four years or so ago suggesting that due to the ncurses license he would be able to deny Thomas Dickey ownership of his own code when Thomas (who had been doing all significant development on ncurses for a year or more) attempted to make his own release of ncurses (4.1, as I recall).

Fortunately all parties agreed to the code being assigned to the FSF, getting ESR out owning a critical-path library.

Mike.

#10 lore raw
In reply to Eric S. Raymond
Quoted 4 lines · Eric S. Raymond

I heard Linus say that he wanted Configure.help split up. I agreed to do it, too, in CML2. But I thought I was not supposed to mess with existing formats -- in fact, I was afraid to, because you and Al Viro beat me up so hard about changing *nothing* before cutover. The CML1 code owns that

Ah, now we see the violence inherent in the system.

Quoted 6 lines

format, and I am not the CML1 maintainer. In fact mec specifically turned me down when I volunteered for that job; he said he wanted me concentrating on CML2.

Neither Linus nor anyone else ever said to me "Eric, it's your job to make that change." When I asked for guidance, Linus blackholed my mail.

Oh! Come and see the violence inherent in the system! Help! Help! I'm being repressed!

Quoted 1 line

Was I supposed to be practicing telepathy, here?

Oh, what a dead giveaway. Did you hear that, did you hear that, eh? That's what I'm on about - did you see him repressing me, you saw it, didn't you?

Dennis^WEric, just one question: had it ever occured to you that
for a self-proclaimed "hacker of social systems" you are doing spectaculary poor job?

#11 lore raw
In reply to Eric S. Raymond

Alexander Viro <[email protected]>:

Quoted 5 lines
Quoted 2 lines

Neither Linus nor anyone else ever said to me "Eric, it's your job to make that change." When I asked for guidance, Linus blackholed my mail.

Oh! Come and see the violence inherent in the system! Help! Help! I'm being repressed!

Your reply was pretty funny. Well, I laughed anyway.

But laughter won't make the chronic communications problems go away.

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#136 lore raw
In reply to Jeff Garzik

Dave Jones <[email protected]>:

Quoted 4 lines

No, but at the least keeping Linux-Kernel in the loop would be considered not just nice, but a somewhat more courteous method than 'sending someone around to go see Linus' when your pet project isn't getting the acceptance you hoped for.

No sending around is involved. Dirk approached me, not the other way around.

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#12 lore raw
In reply to Eric S. Raymond
Quoted 2 lines · In article <[email protected]> you wrote:

But laughter won't make the chronic communications problems go away.

A "I don't like don't like <FOO> and won't apply it" is not a communications problem in general. When the person trying to submit <FOO> keeps submitting it without listening to suggestions or to reasons why it's not applied.... THEN there is a one sided communcation problem.

Not saying that's the case here, but it starts to appear to be so.

#18 lore raw
In reply to Eric S. Raymond

Arjan van de Ven <[email protected]>:

Quoted 3 lines

A "I don't like don't like <FOO> and won't apply it" is not a communications problem in general. When the person trying to submit <FOO> keeps submitting it without listening to suggestions or to reasons why it's not applied.... THEN there is a one sided communcation problem.

Not saying that's the case here, but it starts to appear to be so.

Arjan, I would dearly love to get useful feedback on why things like my Configure.help patches weren't applied. It would be a delight like water in the desert to *hear* some reasons!

But I think you know very well that the usual flow looks like this:

1. You throw a patch over the wall to Linus.

2. Either it shows up in the next release...

3. ...or you hear a vast and echoing silence.

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#27 lore raw
In reply to Eric S. Raymond
Quoted 11 lines · Eric S. Raymond

Arjan, I would dearly love to get useful feedback on why things like my Configure.help patches weren't applied. It would be a delight like water in the desert to *hear* some reasons!

But I think you know very well that the usual flow looks like this:

1. You throw a patch over the wall to Linus.

2. Either it shows up in the next release...

3. ...or you hear a vast and echoing silence.

You're telling me Linus never mailed you about splitting up Configure.help ?

#38 lore raw
In reply to Eric S. Raymond

Arjan van de Ven <[email protected]>:

Quoted 9 lines
Quoted 7 lines

But I think you know very well that the usual flow looks like this:

1. You throw a patch over the wall to Linus.

2. Either it shows up in the next release...

3. ...or you hear a vast and echoing silence.

You're telling me Linus never mailed you about splitting up Configure.help ?

That's right. The only time I ever saw Linus express an opinion on this was in a post on lkml in which he said he didn't like the big monolithic help file. It didn't seem especially directed to me; I am not the CML1 maintainer.

But I took it as a suggestion for CML2. I told him I was willing to do it after CML2 cutover, but was not sure he had considered the ramifications for maintaining rulebase translations. To this request for clarification and guidance, I received no reply.

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#42 lore raw
In reply to Eric S. Raymond
Quoted 5 lines · Eric S. Raymond
Quoted 1 line

You're telling me Linus never mailed you about splitting up Configure.help ?

That's right. The only time I ever saw Linus express an opinion on this was in a post on lkml in which he said he didn't like the big monolithic help file. It didn't seem especially directed to me; I am not the CML1 maintainer.

I recall the threads you mention.
I also recall at least a half dozen regular contributors agreeing that splitting up Configure.help was the way forward.

Quoted 2 lines

But I took it as a suggestion for CML2. I told him I was willing to do it after CML2 cutover

This is something I never understood. CML1 has served us for 10 years or so, and suddenly its declared "unfixable". "I'm not hacking that, I'll fix it in CML2" seemed to become the attitude. End result ? Linus got pissed off with your complaints of non-inclusion, and _fixed_ one of the CML1 problems (Configure.help) And how long did it take ? ISTR it was around an hour after his first mail was sent to you, me & Jeff. 1 hour. Versus the year you spent procrastinating about how CML2 would fix it.

Sure CML2 fixes some bits that are not easily fixed in CML1, but I wonder sometimes how much of it is/was fixable.

Signature · 2 lines
| Dave Jones.        http://www.codemonkey.org.uk
| SuSE Labs
#66 lore raw
In reply to Eric S. Raymond
Quoted 5 lines · Eric S. Raymond

Arjan van de Ven <[email protected]>:

Quoted 4 lines
Quoted 4 lines

But I think you know very well that the usual flow looks like this:

1. You throw a patch over the wall to Linus.

Quoted 7 lines
Quoted 5 lines
Quoted 3 lines

2. Either it shows up in the next release...

3. ...or you hear a vast and echoing silence.

You're telling me Linus never mailed you about splitting up Configure.help ?

That's right.

Gimme a break. I read those posts, I repeatedly saw that people said to do that, I don't remember if it was Linus or not, but it's not like he has the only brain. The clearly expressed sentiment was to put the help next to the source. You repeatedly came up with corner cases where that wouldn't work and used that as an excuse to do nothing. Having just gone through the same thing with Linus about BK locking, I can assure you that dredging up the corner cases does not work, you might as well get to work and solve the problem.

Signature · 2 lines
---
Larry McVoy            	 lm at bitmover.com           http://www.bitmover.com/lm 
#68 lore raw
In reply to Eric S. Raymond

Larry McVoy <[email protected]>:

Quoted 5 lines

Gimme a break. I read those posts, I repeatedly saw that people said to do that, I don't remember if it was Linus or not, but it's not like he has the only brain. The clearly expressed sentiment was to put the help next to the source. You repeatedly came up with corner cases where that wouldn't work and used that as an excuse to do nothing.

And one of those corner cases in fact came around and bit us on the ass.

If we had handled the transition the smooth way I wanted to, and was prepared to, we'd still have a maintainable help database spanning 2.4 as well as 2.5. As it is, Linus blew my system into flinders and obsoleted all my tools (I'm not talking CML2 itself now but the machinery I wrote for tracking help changes).

As a result, I had to tell Marcelo I had no choice but to drop maintaining the 2.4 help file. The divergence, and the damage, is probably not recoverable.

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#72 lore raw
In reply to Eric S. Raymond

Dave Jones <[email protected]>:

Quoted 2 lines

This is something I never understood. CML1 has served us for 10 years or so, and suddenly its declared "unfixable".

Yes, and not by me. By the people who wrote it.

Quoted 10 lines

"I'm not hacking that, I'll fix it in CML2" seemed to become the attitude. End result ? Linus got pissed off with your complaints of non-inclusion, and _fixed_ one of the CML1 problems (Configure.help) And how long did it take ? ISTR it was around an hour after his first mail was sent to you, me & Jeff. 1 hour. Versus the year you spent procrastinating about how CML2 would fix it.

Sure CML2 fixes some bits that are not easily fixed in CML1, but I wonder sometimes how much of it is/was fixable.

The first thing I heard, from mec two years ago, was "the CML1 code base is not salvageable". This was then and is now the unanimous opinion of the kbuild team. Not just mine; in fact, they concluded it before I entered the picture.

I have seen nothing since to make me change that opinion.

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#73 lore raw
In reply to Eric S. Raymond
Quoted 3 lines · Eric S. Raymond

As a result, I had to tell Marcelo I had no choice but to drop maintaining the 2.4 help file. The divergence, and the damage, is probably not recoverable.

find . -name Config.help -exec cat {} >Configure.help \;

Ok, they'll perhaps not be in the same order as your original working set, but that issue goes away when you split your set up with the tools Linus gave you, and regenerate.

Maintaining is hard, lets go shopping.

Signature · 2 lines
| Dave Jones.        http://www.codemonkey.org.uk
| SuSE Labs
#43 lore raw
In reply to Dave Jones
Quoted 2 lines

Sure CML2 fixes some bits that are not easily fixed in CML1, but I wonder sometimes how much of it is/was fixable.

Pretty much all of it. I wrote a proof of concept parser that can deduce the set of rules that must be enforced and what must be changed when you hit an option

#133 lore raw
In reply to Eric S. Raymond
Quoted 11 lines · Eric S. Raymond

Alexander Viro <[email protected]>:

Quoted 6 lines
Quoted 3 lines

Neither Linus nor anyone else ever said to me "Eric, it's your job to make that change." When I asked for guidance, Linus blackholed my mail.

Oh! Come and see the violence inherent in the system! Help! Help! I'm being repressed!

Your reply was pretty funny. Well, I laughed anyway.

But laughter won't make the chronic communications problems go away.

Speaking as a proud member of Al Viro's killfile (from one work account and one home accounts, each time for lkml postings he wasn't cc:'d on, which somehow set him off anyway), I really doubt he perceives lack of communication as much of a problem. More of a solution. :)

Seeing a conspiracy in the existence of any posts to public mailing lists other than linux-kernel is, however, the kind of paranoia for which medication is prescribed. (Would the reactionary front have preferred Eric email the various developers directly in a big CC:, instead of a posting to a PUBLIC LIST, which apparently Jeff DID find and highlighted? I'd say the "help help this is being repressed" came at the start of this thread, not when Eric jumped in.)

And when did talking to Linus directly (in email or on the phone) become something that Linus needed to be protected from? (Look out guys! Some people are thinking of TALKING DIRECTLY TO LINUS! Without consulting us first! They must be found and stopped!)

And here I thought Linus was very good at saying no, in a multitude of contexts...

Rob

#74 lore raw
In reply to Eric S. Raymond

Dave Jones <[email protected]>:

Quoted 5 lines
Quoted 3 lines

As a result, I had to tell Marcelo I had no choice but to drop maintaining the 2.4 help file. The divergence, and the damage, is probably not recoverable.

find . -name Config.help -exec cat {} >Configure.help \;

Back-seat drivers. Answers for everything, understanding of nothing.

Sorry, it's not *nearly* that simple. For one thing, one of the fourteen billion requirements I've had dropped on me is that Marcelo doesn't want any symbols in his help file that aren't actually in the 2.4 rulebase.

So I'd have to generate a custom version for him. I used to have the machinery to do that by parsing magic comments in my master file. Until Linus blew it apart.

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#44 lore raw
In reply to Dave Jones

Alan Cox <[email protected]>:

Quoted 6 lines
Quoted 2 lines

Sure CML2 fixes some bits that are not easily fixed in CML1, but I wonder sometimes how much of it is/was fixable.

Pretty much all of it. I wrote a proof of concept parser that can deduce the set of rules that must be enforced and what must be changed when you hit an option

Alan. It didn't work. It couldn't have -- among other things, the old language can't tell visibility from implication.

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#45 lore raw
Quoted 6 lines
Quoted 3 lines

Pretty much all of it. I wrote a proof of concept parser that can deduce the set of rules that must be enforced and what must be changed when you hit an option

Alan. It didn't work. It couldn't have -- among other things, the old language can't tell visibility from implication.

The prototype generates a very convincing table set, and the tables generate very convincing graphs. The information to work out what option needs another as I've said for months

#76 lore raw
In reply to Eric S. Raymond
Quoted 1 line · Eric S. Raymond

Back-seat drivers. Answers for everything, understanding of nothing.

*plonk*

Quoted 3 lines

Sorry, it's not *nearly* that simple. For one thing, one of the fourteen billion requirements I've had dropped on me is that Marcelo doesn't want any symbols in his help file that aren't actually in the 2.4 rulebase.

As you obviously completely ignored my previous point, I'll reiterate it.

1. You run Linus' split-into-config.in script on a 2.4 tree.
2. You add whatever Config.ins have been updated to the 2.4 config.ins
3. You regenerate with the find example I showed in the other mail.

There. 2.4 Configure.help up to date with latest symbols, but containing none of the 2.5 ones.

Signature · 2 lines
| Dave Jones.        http://www.codemonkey.org.uk
| SuSE Labs
#58 lore raw
In reply to Dave Jones
Quoted 9 lines · Dave Jones

This is something I never understood. CML1 has served us for 10 years or so, and suddenly its declared "unfixable". "I'm not hacking that, I'll fix it in CML2" seemed to become the attitude. End result ? Linus got pissed off with your complaints of non-inclusion, and _fixed_ one of the CML1 problems (Configure.help) And how long did it take ? ISTR it was around an hour after his first mail was sent to you, me & Jeff. 1 hour. Versus the year you spent procrastinating about how CML2 would fix it.

Amen. This whole "CML1 vs CML2" thing is silly. We don't see other "maintainers" refusing to touch code, having no relation to code, and then coming out of no-where and offering some external solution for the perceived problem.

If you _develop_ kernel code and see a problem with the configure system, then fix the problems. As Linus says, let stuff evolve.

Quoted 2 lines

Sure CML2 fixes some bits that are not easily fixed in CML1, but I wonder sometimes how much of it is/was fixable.

Agreed, but CML2 has a whole bunch of new "features" too which will be shoved down our throats!!

Robert Love

#75 lore raw
In reply to Eric S. Raymond
Quoted 3 lines · Eric S. Raymond

So I'd have to generate a custom version for him. I used to have the machinery to do that by parsing magic comments in my master file. Until Linus blew it apart.

Repeat after me: Linus is a bastard. Linus doesn't care.

Everyone knows this. Even Linus admits he's a bastard. We all have to live with changes. It's frustrating when some code you're maintaining code for different kernel trees and you can't share a common master due to API changes. But Linus doesn't care. This is the price of doing business here. Everyone knows that. And (almost) everyone keeps coming back for more punishment. People live with the system.

So, can we please drop this thread? You're frustrated. Fine. You've had your chance to vent. OK. Everybody needs to do that from time to time. But move on. Don't keep flogging this dead horse. You're cutting into the bone, now.

				Regards,

					Richard....
Permanent: [email protected]
Current:   [email protected]
#46 lore raw
In reply to Eric S. Raymond

Alan Cox <[email protected]>:

Quoted 3 lines

The prototype generates a very convincing table set, and the tables generate very convincing graphs. The information to work out what option needs another as I've said for months

I've *solved* this problem. I understand the constraints. I know exactly what you will have to do to get the rest of the way, *because I did it*.

I have built not just a "proof of concept" but a working implementation so robust that it is in production use by three different projects. And a substantial number of kernel developers are using it now and reporting it good.

The good thing about working with developers like you and Dave Jones and many of the other loonies on this list is that you are fucking brilliant. The bad thing is that because you're fucking brilliant, you're often also ungodly arrogant about second-guessing other peoples' work -- you think your snap judgment or "proof of concept" is somehow equivalent to two years of design, coding and testing by someone who has been concentrating on the problem.

News bulletin: IT'S NOT.

Alan, don't talk to me about "proof of concept". Tell me about a production-quality system, proven in use by people like Embedsys, Webmachines, and the Compache project. Tell me you can duplicate what CML2 does successfully before you run around implying my design assumptions are full of crap.

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#77 lore raw
In reply to Eric S. Raymond

Richard Gooch <[email protected]>:

Quoted 1 line

Repeat after me: Linus is a bastard. Linus doesn't care.

Fine. We all know Linus is a bastard.

If that's so, then why are the likes of Jeff Garzik and Al Viro spending so much effort trying to make *me* into the bad guy?

Time for somebody to up Jeff's Thorazine dosage...

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#86 lore raw
In reply to Eric S. Raymond

You're not going to get sympathy from this list for having your work "blown apart" by Linus. Fix the problems that are holding things up or give up on it. The current system, CML1, does function so you have to accept you're pushing CML2 up a hill.

} So I'd have to generate a custom version for him. I used to have the } machinery to do that by parsing magic comments in my master file. } Until Linus blew it apart.

#69 lore raw
In reply to Eric S. Raymond
Quoted 3 lines · Eric S. Raymond

As a result, I had to tell Marcelo I had no choice but to drop maintaining the 2.4 help file. The divergence, and the damage, is probably not recoverable.

Divergence and damage to a configure help file ?

As someone who professes to be well-versed in social systems and their associated mechanics, I am sure you understand the notions of social contracts and the like. Thus, if we, as the developers who actually look at the Config.help, drop the changes then that is our problem. And if we don't care, then there must be nothing to care about.

If Marcelo and Linus are not begging for Configure.help updates (like I beg for SCSI layer rewrites) then there is a reason.

Robert Love

#47 lore raw
In reply to Eric S. Raymond
Quoted 7 lines
Quoted 3 lines

The prototype generates a very convincing table set, and the tables generate very convincing graphs. The information to work out what option needs another as I've said for months

I've *solved* this problem. I understand the constraints. I know exactly what you will have to do to get the rest of the way, *because I did it*.

No you've replaced the system with one thats much harder to understand and forces new tools on people. The *interesting* problem to solve is keeping the existing infrastructure there and getting the same kind of results.

Since the information is there in CML1 to generate the list of constraints for any given option, its a reasonable assertion that the entire CML2 language rewrite is self indulgence from a self confessed language invention freak.

Alan

#13 lore raw
In reply to Arjan van de Ven
Quoted 2 lines · Eric S. Raymond

If that's so, then why are the likes of Jeff Garzik and Al Viro spending so much effort trying to make *me* into the bad guy?

'the bad guy' probably isn't the right choice of words. The point being made is that you constantly refuse to listen. You ask for an opinion, and when something comes back that isn't what you expect, you run off and write something completely different just to appease some fabled Aunt Tilley, then come back and say "Ok, how about now?".

CML2 seems to have all these wonderous features to make end-users happy. In writing these, you've shifted target audience from kernel _developers_ to kernel _users_. Concerns from developers were constantly ignored in favour of adding extra candy. And you wonder why developers don't want to read about CML2 any more?

You know, for all its glorious features, I'd not be surprised to see mconfig or some other 'improved CML1' end up in 2.6 rather than CML2.

Quoted 1 line

Time for somebody to up Jeff's Thorazine dosage...

Increased dosages of something all around wouldn't go amiss I think.

Signature · 2 lines
| Dave Jones.        http://www.codemonkey.org.uk
| SuSE Labs
#70 lore raw
In reply to Robert Love

The amount of whining I've had to read today on this topic just astonishes me. People, please - devote 1/10th of the energy to helping esr finish and fix CML2 to meet Linus' expectations. Face the fact: CML1 was good for it's time, but that time is now gone.

If you spent as much time helping to get CML2 up and going as some of you have trying to kill it (while not offering anything constructive to replace it) it would have been DONE 6 months ago.

Sometimes, this list is worse than the proverbial sewing circle.

Quoted 1 line
Quoted 1 line

As a result, I had to tell Marcelo I had no choice but to drop

maintaining

Quoted 22 lines
Quoted 2 lines

the 2.4 help file. The divergence, and the damage, is probably not recoverable.

Divergence and damage to a configure help file ?

As someone who professes to be well-versed in social systems and their associated mechanics, I am sure you understand the notions of social contracts and the like. Thus, if we, as the developers who actually look at the Config.help, drop the changes then that is our problem. And if we don't care, then there must be nothing to care about.

If Marcelo and Linus are not begging for Configure.help updates (like I beg for SCSI layer rewrites) then there is a reason.

Robert Love

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

#87 lore raw
In reply to Eric S. Raymond
Quoted 6 lines · Richard Gooch
Quoted 3 lines · Eric S. Raymond

So I'd have to generate a custom version for him. I used to have the machinery to do that by parsing magic comments in my master file. Until Linus blew it apart.

Repeat after me: Linus is a bastard. Linus doesn't care.

And he owes me $5 which he won't pay

Signature · 4 lines
---------------------------------------------------------
Victor Yodaiken 
Finite State Machine Labs: The RTLinux Company.
 www.fsmlabs.com  www.rtlinux.com
#78 lore raw
In reply to Eric S. Raymond
Quoted 2 lines

If that's so, then why are the likes of Jeff Garzik and Al Viro spending so much effort trying to make *me* into the bad guy?

I don't think they are. They are just watching you do it yourself

Quoted 1 line

Time for somebody to up Jeff's Thorazine dosage...

See...

#88 lore raw
In reply to Eric S. Raymond
Quoted 4 lines · Robert Love
Quoted 2 lines

Sure CML2 fixes some bits that are not easily fixed in CML1, but I wonder sometimes how much of it is/was fixable.

Agreed, but CML2 has a whole bunch of new "features" too which will be shoved down our throats!!

Increased functionality I don't have a problem with, as long as other more important things are addressed. And for that matter, Linus has said to Eric "I don't care, take this out of the kernel completely leaving just oldconfig'.

It might just be the sanest choice, and leave all this discussion in userland.

Signature · 2 lines
| Dave Jones.        http://www.codemonkey.org.uk
| SuSE Labs
#90 lore raw
In reply to Eric S. Raymond

Dave Jones <[email protected]>:

Quoted 8 lines

As you obviously completely ignored my previous point, I'll reiterate it.

1. You run Linus' split-into-config.in script on a 2.4 tree.
2. You add whatever Config.ins have been updated to the 2.4 config.ins
3. You regenerate with the find example I showed in the other mail.

There. 2.4 Configure.help up to date with latest symbols, but containing none of the 2.5 ones.

And you've completely ignored the real problem...which is when I get a text change for one tree, *how do I automatically propagate it to the other*? How do I *tell* that it ought to be propagated?

It's not sufficient to develop a one-shot conversion procedure. What I came up with had to allow me to continue generating correct help files for all trees under the assumption that text changes for 2.4 could come in against the 2.5 copy of the help entry, or vice-versa -- bearing in mind that some symbols have divergent text and that's correct.

Solutions that involve me doing an arbitrary and increasing amount of hand-hacking on every release are right out.

If you think this problem through, I'm sure you'll come up with a design very similar to what I actually built. Which, by Linus's choice, got irrecoverably nuked.

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#52 lore raw
In reply to Eric S. Raymond

Alan Cox <[email protected]>:

Quoted 2 lines

Since the information is there in CML1 to generate the list of constraints for any given option,

False assumption...

Quoted 3 lines

its a reasonable assertion that the entire CML2
language rewrite is self indulgence from a self confessed language invention freak.

leading to a false result.

If you want to refute me, build it yourself. You'll get a valuable learning experience. At the end of it, I'll have earned your apology. Not that I expect to get it.

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#91 lore raw
In reply to Eric S. Raymond

Dave Jones <[email protected]>:

Quoted 7 lines

Increased functionality I don't have a problem with, as long as other more important things are addressed. And for that matter, Linus has said to Eric "I don't care, take this out of the kernel completely leaving just oldconfig'.

It might just be the sanest choice, and leave all this discussion in userland.

CML2 can live in userland. It already does; three other projects use it. The kernel rulebase cannot. The real issue here is whether the CML1 language carries sufficient information to support things like side-effect deduction. And the answer is "no".

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#14 lore raw
In reply to Arjan van de Ven
Quoted 7 lines · Eric S. Raymond

Richard Gooch <[email protected]>:

Quoted 1 line

Repeat after me: Linus is a bastard. Linus doesn't care.

Fine. We all know Linus is a bastard.

If that's so, then why are the likes of Jeff Garzik and Al Viro spending so much effort trying to make *me* into the bad guy?

Perhaps they don't like the results of your efforts and they resent what they perceive to be end runs around the normal process of getting changes accepted. As well they should. Your attempt to avoid peer review is disgusting, it's the worst of what happens in closed source shops when marketing forces the wrong answer in over the objections of the people who have to live with it. Yuck.

If you really want to contribute, what you'll do is improve the existing system. Screaming that it can't be done just means you aren't a kernel programmer. What kernel programmers do is the hard ugly work it takes to make things work smoothly. Look at any device driver - just a handfull of simple interfaces: open,close,read,write,ioctl (ok that last one is far from simple). Now go look at the code implementing those interfaces, it's frequently a huge effort to make some recalcitrant device behave.

But that's what kernel programmers do. They take a mess and present a clean, well working system. No one said it was easy. If you are a kernel programmer, you can take CML1 and make it work, for a reasonable definition of "work". If you can't, you're far better off to admit you're in over your head and give it up. Harping on CML2 as the better answer isn't going to work. Neither is telling people they don't get it, when those same people have done 10,000 times more work on the kernel than you'll ever do.

Signature · 2 lines
---
Larry McVoy            	 lm at bitmover.com           http://www.bitmover.com/lm 
#53 lore raw
In reply to Eric S. Raymond
Quoted 5 lines

Alan Cox <[email protected]>:

Quoted 2 lines

Since the information is there in CML1 to generate the list of constraints for any given option,

False assumption...

Do you have anything more constructive that "doesnt" to say

Quoted 3 lines

If you want to refute me, build it yourself. You'll get a valuable learning experience. At the end of it, I'll have earned your apology. Not that I expect to get it.

Been there, done that, got the pretty graphs. Possibly the next step is to redo the work into mconfig which already does pretty much anything we need with the existing config files parsed using a real 100% genuine yacc grammar

#15 lore raw
In reply to Arjan van de Ven
Quoted 3 lines · Eric S. Raymond

And you've completely ignored the real problem...which is when I get a text change for one tree, *how do I automatically propagate it to the other*? How do I *tell* that it ought to be propagated?

Depends on the context of the change.
In a majority of cases the answer is RTFC.
If this is too much effort, rethink your outlook on whats involved as 'maintainer' of something like the Config.help's.

This is precisely the reason that splitting them was the best thing that could have happened. They are now maintained by the people who actually *know* whats involved, so you don't have to.

Quoted 2 lines

Solutions that involve me doing an arbitrary and increasing amount of hand-hacking on every release are right out.

Reading diffs and finding out why things changed, and following conversations on Linux-kernel are the only way you'll know if something is relevant in other trees. If you've a robot that can do this, I'm all ears, as it could save me hours each day.

Signature · 2 lines
| Dave Jones.        http://www.codemonkey.org.uk
| SuSE Labs
#55 lore raw
In reply to Eric S. Raymond
Quoted 5 lines · Eric S. Raymond

Alan, don't talk to me about "proof of concept". Tell me about a production-quality system, proven in use by people like Embedsys, Webmachines, and the Compache project. Tell me you can duplicate what CML2 does successfully before you run around implying my design assumptions are full of crap.

Eric, step back a sec. Deep breaths.

Nobody ever said CML1 had to be able to serve projects other than linux-kernel. The fact CML2 can is nice, but irrelevant. That argument goes nowhere.

The amount of time you've invested in your code isn't particularly interesting to anybody but you, except as an unreliable yardstick of how long it might take to duplicate things. The "genius from mars" technique might be able to come up with an even better way in a week, you never know. (Even a really hard problem space can be elegantly solved out of left field. It's not likely, but you never know.)

The other side of the argument is that a proof of concept is not the same thing as actually solving the problem. You have working code. Several other people have unimplemented theoretical proposals, tests, and prototypes that don't actually do anything useful as of yet. That's the main argument you should probably be making.

If people want to prove Eric wrong, then make CML1 work well. If you believe it's not as hard a problem as he says it is, then by all means prove him wrong. Arguing about how hard the problem really would be to solve, without actually solving it... Why?

Rob

#19 lore raw
In reply to Eric S. Raymond

Dave Jones <[email protected]>:

Quoted 5 lines

The point being made is that you constantly refuse to listen. You ask for an opinion, and when something comes back that isn't what you expect, you run off and write something completely different just to appease some fabled Aunt Tilley, then come back and say "Ok, how about now?".

I honestly don't undestand how this myth got started, Dave.

Quoted 1 line

From my point of view, I have been busting my ass trying to *extract*

requirements from Linus and other people with a stake. What guidance I get is vague and contradictory. My proposals to address it get ignored. Everything I do right gets written off and teensy side issues get inflated into monsters.

It's utterly maddening to hear people accuse me of not listening when I spend so damn much effort trying to do what they want for so little reward...

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#24 lore raw
In reply to Eric S. Raymond

Larry McVoy <[email protected]>:

Quoted 3 lines

Perhaps they don't like the results of your efforts and they resent what they perceive to be end runs around the normal process of getting changes accepted. As well they should. Your attempt to avoid peer review is

                                                   ^^^^^^^^^^^^^^^^^
Quoted 3 lines

disgusting, it's the worst of what happens in closed source shops when marketing forces the wrong answer in over the objections of the people who have to live with it. Yuck.

I've "avoided peer review" really effectively, yeah. Unanimous support from the kbuild team. Three other projects using the software. Dozens of beta testers. Nope, no review at all there. Ain't I clever?

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#92 lore raw
In reply to Eric S. Raymond
Quoted 3 lines

The kernel rulebase cannot. The real issue here is whether the CML1 language carries sufficient information to support things like side-effect deduction. And the answer is "no".

Thats an opinion not an answer. What doesn't it contain ?

#25 lore raw
In reply to Eric S. Raymond
Quoted 8 lines · Larry McVoy

If you really want to contribute, what you'll do is improve the existing system. Screaming that it can't be done just means you aren't a kernel programmer. What kernel programmers do is the hard ugly work it takes to make things work smoothly. Look at any device driver - just a handfull of simple interfaces: open,close,read,write,ioctl (ok that last one is far from simple). Now go look at the code implementing those interfaces, it's frequently a huge effort to make some recalcitrant device behave.

Quoted 8 lines

But that's what kernel programmers do. They take a mess and present a clean, well working system. No one said it was easy. If you are a kernel programmer, you can take CML1 and make it work, for a reasonable definition of "work". If you can't, you're far better off to admit you're in over your head and give it up. Harping on CML2 as the better answer isn't going to work. Neither is telling people they don't get it, when those same people have done 10,000 times more work on the kernel than you'll ever do.

(cc: list killed, I assume all are subscribed to l-k)

Preface: I have no plan of kbuild and kernel configuration, so take this with a grain of salt. But maybe it gives me the chance to calm all this because I'm not biased towards either side.

Are you telling that kernel programmers don't rewrite code from scratch? Is that a correct interpretation of "improve the existing system"? Note that "it can't be done" can also imply "cannot reasonable be done".

If that's not what you mean, stop reading this mail, drop a line to clarify this and forget this piece of mail. If it _is_ what you mean, then with all due respect, why cannot kernel programmers rewrite a subsystem (even if it interfaces with the whole world of kernel configurations) from scratch?

Eric has done it, without being of kernel hacker temple's fame.

Whether he doesn't listen to other developers, I cannot tell, I got my /fetchmail/ issues resolved. Still, I share the opinion that the kernel build stuff is mainly for developers (and if it cuts down turnaround times, it well deserves a try), and if -- as a side effect -- it's good for the user, so be it, and if it's clear and self documenting, that's nothing a developer would reject. Resurrecting this IMHO dead thread around Mrs. Tillie won't bring any good. (Although Aunt Tillie should really get her kernel as .deb or .rpm.)

AFAICS, Eric has gone lengths to get a competitive kernel configuration system working well enough for production use, minor issues seem to remain, like split Configure.help. I can easily understand no-one is throwing this work away. If he's convinced the system is more consistent and robust in itself, that's good -- but all this advocacy seems to be torn down to some (personal) vendetta. Linux does not deserve that.

What I'm missing in this thread are facts. The point "communications problem" has been raised. It seems to me as though the two opposing parties (pro and con CML2) were not clear about what they expect from the other party. If something has gone on behind the scenes, well, I will have missed it...

I'm really hoping that this issue can get resolved in one way or another, my personal opinion is that one single config stuff parser is the goal, be it mconfig or be it CML2.

#99 lore raw
In reply to Eric S. Raymond
Quoted 2 lines · Eric S. Raymond

Solutions that involve me doing an arbitrary and increasing amount of hand-hacking on every release are right out.

Um, Eric? Isn't that what being a maintainer basically means?

Okay, Linus's changes killed backwards compatability with 2.4. (He does that. That's what 2.5 is for. It would be nice if less of it happened in the stable series, but... :)

A maintainer is basically there to serve Linus (the project's undisputed architect), who periodically makes the architectural decision to break backwards compatability on some front.

You wanted to remain 2.4 help file maintainer as well as 2.5, and Linus did not address this concern (for whatever reason: Linus blackholing email as a way of disagreeing with it led to a lot of miscommunications like this one.) Fine. Water under the bridge. Linus made an architectural decision, and it has been implemented. If it means you can't maintain 2.4 anymore, the sane move is to drop 2.4 (which you did) and move on.

That is why this is a dead issue, and bringing it up again doesn't serve as much purpose as you think it does. (If you want to highlight the blackhole-related miscommunication, fine. But even implying Linus's architectural decision was wrong carries no weight with your audience.)

Quoted 3 lines

If you think this problem through, I'm sure you'll come up with a design very similar to what I actually built. Which, by Linus's choice, got irrecoverably nuked.

Tough.

Eric, I like you and I still don't agree with you on this one.

First of all, the help files really are largely orthogonal to CML2. They could be written in gzipped ebcdic. Make a filter, read them, move on. You are pointelessly wasting brownie points on a dead side issue.

Secondly, please define the problem space you're going after. (I think that the main objection people have to this tool, they don't agree with the definition of the problem you're trying to solve.)

If you want CML2 is to be the best configure tool for the 2.5 (and later) kernel series, fine. Then FOCUS ON THAT PROBLEM.

Overhead to deal with 2.4 is bloat. Flexibility for projects outside of the kernel itself is a side issue. Aunt Tillie is NOT the initial target audience. Usage to configure debian userspace or some such is completely tangential. All of the above may be fun, but they do not serve to advance the primary objective, and bringing them up does not help make a case for CML2.

Back up a bit. What would be the most minimal, stripped-down version of CML2 you could write? No eye candy, no complications, no autoconfigurator, no tree view, no frozen symbols. Just solving the core problem of configuring 2.5 in a more flexible and less buggy way than CML1, with the three interfaces (oldconfig, menuconfig, xconfig) we've got now.

Think about that for a while. Try to get THAT into the kernel. THEN worry about building on top of that.

Remember: Linus likes small patches. (CONCEPTUALLY small if possible, not just lines of code...)

Rob

#28 lore raw
In reply to Arjan van de Ven
Quoted 4 lines · Matthias Andree

Are you telling that kernel programmers don't rewrite code from scratch? Is that a correct interpretation of "improve the existing system"? Note that "it can't be done" can also imply "cannot reasonable be done". Eric has done it, without being of kernel hacker temple's fame.

The kernel hacker approach: Gradual change toward a predefined goal. The Eric approach: Rip out existing, replace with new.

If Al Viro can rewrite the guts of the VFS without hardly anyone noticing any disturbance, and the configuration system can't be done this way, something is amiss.

Signature · 2 lines
| Dave Jones.        http://www.codemonkey.org.uk
| SuSE Labs
#30 lore raw
In reply to Arjan van de Ven
Quoted 6 lines · Matthias Andree

Are you telling that kernel programmers don't rewrite code from scratch? Is that a correct interpretation of "improve the existing system"? Note that "it can't be done" can also imply "cannot reasonable be done".

If that's not what you mean, stop reading this mail, drop a line to clarify this and forget this piece of mail.

That's not what I mean. But it's worthwhile to note that almost all "rewrite from scratch" projects really translate into "I'm unwilling to learn what the last guy did, and I'm smarter, so I'm going to do what I want to do". And that is not what kernel programmers do. They would love to be able to do that but it's very rare that doing so makes sense.

In reality, the guy who came before you wasn't an idiot. Maybe s/he didn't do the best job possible, but the fact that the code is there and works implies some level of understanding. Real programmers make their work fit *seamlessly* with the work of the people who came before them. It's egoless programming, you are striving to make things fit so well that noone can see where person A stopped and person B started.

In Eric's case, it would have been far better if he had done some work to make CML1 work better. It simply isn't broken enough to justify throwing it out. And it's not at all clear to me, at least, that the CML2 stuff is any less broken, it's just broken in different ways. Unless Eric can make a system that is widely viewed as better overall than CML1, making one that is better in some ways but worse or just as bad in others, well, that's a big waste of time.

At Sun, we had a rule that a major change was automatically disallowed unless it showed a factor of 2 improvement in some important way while holding the other variables constant. I don't think CML2 would pass that test. Aunt Tillie's opinion doesn't count.

Signature · 2 lines
---
Larry McVoy            	 lm at bitmover.com           http://www.bitmover.com/lm 
#79 lore raw
In reply to Eric S. Raymond
Quoted 7 lines · Eric S. Raymond

Richard Gooch <[email protected]>:

Quoted 1 line

Repeat after me: Linus is a bastard. Linus doesn't care.

Fine. We all know Linus is a bastard.

If that's so, then why are the likes of Jeff Garzik and Al Viro spending so much effort trying to make *me* into the bad guy?

I can't speak for Jeff. Compared to me Linus is downright nice, kind and touchy-feely guy. I'm not. I don't like manipulative arseholes. I _despise_ inept manipulative arseholes and I really can't stand demagogy - what with a massive overdose of that in school years (Soviet Union, early 80s - 'nuff said).

Your habit of claiming completely unsubstantiated credentials also doesn't help. Let me spell it out - your pseudo-phylosophical essays are handwaving at best and deliberate dishonesty at worst. If you expect respect for them - you are looking in the wrong place. You don't understand how kernel development works - that much had been made obvious for everyone by the last couple of months. And yet you don't hesitate to plug holes in your proposals with empty preaching ex cathedra and massive demagogy.

I had seen such guys in Uni. Usually they taught Marxism-Leninism, History of Party, Philosophy, etc. Nobody with a shred of clue and self-respect had touched them with a ten-feet pole. Excuse me for being not too happy to see that type again.

#29 lore raw
In reply to Dave Jones

by the way folks, unless Eric is plain lying to us the 'private flamewar' that Linus has with someone that prompted him to splitup the config.help stuff was not with Eric.

David Lang

Quoted 27 lines · Dave Jones

Date: Sat, 16 Feb 2002 00:29:59 +0100
From: Dave Jones <[email protected]>
To: [email protected]
Subject: Re: Disgusted with kbuild developers

Quoted 4 lines · Matthias Andree

Are you telling that kernel programmers don't rewrite code from scratch? Is that a correct interpretation of "improve the existing system"? Note that "it can't be done" can also imply "cannot reasonable be done". Eric has done it, without being of kernel hacker temple's fame.

The kernel hacker approach: Gradual change toward a predefined goal. The Eric approach: Rip out existing, replace with new.

If Al Viro can rewrite the guts of the VFS without hardly anyone noticing any disturbance, and the configuration system can't be done this way, something is amiss.

--
| Dave Jones. http://www.codemonkey.org.uk
| SuSE Labs
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

#31 lore raw
In reply to Larry McVoy

I wasn't at the kernel summit meeting, but my undersanding from the 'press' reports was that it was the CML1 guys who said they wanted to throw it out and start from scratch. I don't know what role Eric had in the discussion, but it sure didn't sound like it was him alone prompting the change.

David Lang

Quoted 47 lines · Larry McVoy

Date: Fri, 15 Feb 2002 15:36:36 -0800
From: Larry McVoy <[email protected]>
To: [email protected]
Subject: Re: Disgusted with kbuild developers

Quoted 6 lines · Matthias Andree

Are you telling that kernel programmers don't rewrite code from scratch? Is that a correct interpretation of "improve the existing system"? Note that "it can't be done" can also imply "cannot reasonable be done".

If that's not what you mean, stop reading this mail, drop a line to clarify this and forget this piece of mail.

That's not what I mean. But it's worthwhile to note that almost all "rewrite from scratch" projects really translate into "I'm unwilling to learn what the last guy did, and I'm smarter, so I'm going to do what I want to do". And that is not what kernel programmers do. They would love to be able to do that but it's very rare that doing so makes sense.

In reality, the guy who came before you wasn't an idiot. Maybe s/he didn't do the best job possible, but the fact that the code is there and works implies some level of understanding. Real programmers make their work fit *seamlessly* with the work of the people who came before them. It's egoless programming, you are striving to make things fit so well that noone can see where person A stopped and person B started.

In Eric's case, it would have been far better if he had done some work to make CML1 work better. It simply isn't broken enough to justify throwing it out. And it's not at all clear to me, at least, that the CML2 stuff is any less broken, it's just broken in different ways. Unless Eric can make a system that is widely viewed as better overall than CML1, making one that is better in some ways but worse or just as bad in others, well, that's a big waste of time.

At Sun, we had a rule that a major change was automatically disallowed unless it showed a factor of 2 improvement in some important way while holding the other variables constant. I don't think CML2 would pass that test. Aunt Tillie's opinion doesn't count.
--
---
Larry McVoy lm at bitmover.com http://www.bitmover.com/lm
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

#35 lore raw
In reply to Arjan van de Ven
Quoted 10 lines · Larry McVoy
Quoted 7 lines · Matthias Andree

Are you telling that kernel programmers don't rewrite code from scratch? Is that a correct interpretation of "improve the existing system"? Note that "it can't be done" can also imply "cannot reasonable be done".

If that's not what you mean, stop reading this mail, drop a line to clarify this and forget this piece of mail.

Quoted 5 lines

That's not what I mean. But it's worthwhile to note that almost all "rewrite from scratch" projects really translate into "I'm unwilling to learn what the last guy did, and I'm smarter, so I'm going to do what I want to do". And that is not what kernel programmers do. They would love to be able to do that but it's very rare that doing so makes sense.

I have to call bullshit on this one! :)

The kernel is always getting some module replaced by some other module. Everything from the VM to the recent ALSA merge that will kill OSS. This is not to claim it's always a good thing, just that you can't claim this is not what kernel programmers do...

Hell, we gotta be some of the most egotistical folks around, it's what keeps us going!

Signature · 3 lines
Ben Greear <[email protected]>       <Ben_Greear AT excite.com>
President of Candela Technologies Inc      http://www.candelatech.com
ScryMUD:  http://scry.wanfear.com     http://scry.wanfear.com/~greear
#20 lore raw
In reply to Eric S. Raymond
Quoted 10 lines · Eric S. Raymond

Dave Jones <[email protected]>:

Quoted 5 lines

The point being made is that you constantly refuse to listen. You ask for an opinion, and when something comes back that isn't what you expect, you run off and write something completely different just to appease some fabled Aunt Tilley, then come back and say "Ok, how about now?".

I honestly don't undestand how this myth got started, Dave.

Quoted 1 line

From my point of view, I have been busting my ass trying to *extract*

Quoted 8 lines

requirements from Linus and other people with a stake. What guidance I get is vague and contradictory. My proposals to address it get ignored. Everything I do right gets written off and teensy side issues get inflated into monsters.

It's utterly maddening to hear people accuse me of not listening when I spend so damn much effort trying to do what they want for so little reward...

Eric....

Again... Just like I said to you in private email...

Show us that you're able to write a 1 for 1 functional correspondance between CML1 and CML2 and propose that for inclusion into 2.5. This will already have the advantage of using a common engine to parse the config language regardless of the frontend used, along with the readability and compactness of CML2. This should not cause a too big cultural change.

Wait for a month or two while developers get used to it. Then and only then could you start discussion about advanced CML2 possibilities and features. REmember that Linux development only works like a car that you propose individual parts individually and let people comment on them. Providing the whole car already assembled with a "turn key" package simply doesn't work with most car hackers.

As a bonus to further stimulate acceptance of CML2, make it work with the tools that most people already have i.e. Python 1.5.

Nicolas

#32 lore raw
In reply to David Lang
Quoted 4 lines

'press' reports was that it was the CML1 guys who said they wanted to throw it out and start from scratch. I don't know what role Eric had in the discussion, but it sure didn't sound like it was him alone prompting the change.

I was at the summit but I guess I missed any real discussion on it, other than the out of meeting sequence with Linus basically yelling at Eric over some CML2 stuff.

I still hold the opinion the CML1 declarations are sufficient, and that mconfig is more than enough to do what we need, possibly coupled with some automatic rule validation stuff.

Michael Chastain fixed the nasty parsing mess in 1998

#80 lore raw
In reply to Alexander Viro

Here's an idea Al: Put up, or shut up. Where's your code to replace CML1 with? I like Eric's system, and find it's MUCH more useful in a production environment than CML1 was. Where's your replacement idea?

Scott

----- Original Message -----
From: "Alexander Viro" <[email protected]>
To: "Eric S. Raymond" <[email protected]>
Cc: "Richard Gooch" <[email protected]>; "Dave Jones" <[email protected]>; "Larry McVoy" <[email protected]>; "Arjan van de Ven" <[email protected]>; <[email protected]>
Sent: Friday, February 15, 2002 5:37 PM
Subject: Re: Disgusted with kbuild developers

Quoted 38 lines · Eric S. Raymond
Quoted 8 lines

Richard Gooch <[email protected]>:

Quoted 1 line

Repeat after me: Linus is a bastard. Linus doesn't care.

Fine. We all know Linus is a bastard.

If that's so, then why are the likes of Jeff Garzik and Al Viro spending so much effort trying to make *me* into the bad guy?

I can't speak for Jeff. Compared to me Linus is downright nice, kind and touchy-feely guy. I'm not. I don't like manipulative arseholes. I _despise_ inept manipulative arseholes and I really can't stand demagogy - what with a massive overdose of that in school years (Soviet Union, early 80s - 'nuff said).

Your habit of claiming completely unsubstantiated credentials also doesn't help. Let me spell it out - your pseudo-phylosophical essays are handwaving at best and deliberate dishonesty at worst. If you expect respect for them - you are looking in the wrong place. You don't understand how kernel development works - that much had been made obvious for everyone by the last couple of months. And yet you don't hesitate to plug holes in your proposals with empty preaching ex cathedra and massive demagogy.

I had seen such guys in Uni. Usually they taught Marxism-Leninism, History of Party, Philosophy, etc. Nobody with a shred of clue and self-respect had touched them with a ten-feet pole. Excuse me for being not too happy to see that type again.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

#81 lore raw
In reply to Alexander Viro
Quoted 3 lines · William Scott Lockwood III

Here's an idea Al: Put up, or shut up. Where's your code to replace CML1 with? I like Eric's system, and find it's MUCH more useful in a production environment than CML1 was. Where's your replacement idea?

I think Al's work on the kernel speaks for itself. He's been putting up just fine, thank you. And I think you're looking in the wrong place, Al has expressed no interest (that I'm aware of) in rewriting this code. As a user of the new system, a pretty typical user, his opinion counts.

I don't think that you're really getting the point. Nobody is saying that CML1 is the greatest thing since sliced bread. What they are saying is that it seems to work pretty well, yes it could be better, but CML2 isn't shaping up to be an improvement so much as a Eric Raymond Language exercise. Noone begrudges Eric his right to come up with as many little languages as he wants. But when he asks the kernel developers to use them, he'd better be prepared to hear each and every thing they find wanting, and address the majority of those issues. That hasn't happened. Instead, there have been a lot of flame wars, politics, protests about Linus, etc.

Signature · 2 lines
---
Larry McVoy            	 lm at bitmover.com           http://www.bitmover.com/lm 
#39 lore raw
In reply to Eric S. Raymond
Quoted 6 lines · Dave Jones
Quoted 4 lines · Matthias Andree

Are you telling that kernel programmers don't rewrite code from scratch? Is that a correct interpretation of "improve the existing system"? Note that "it can't be done" can also imply "cannot reasonable be done". Eric has done it, without being of kernel hacker temple's fame.

Quoted 6 lines

The kernel hacker approach: Gradual change toward a predefined goal. The Eric approach: Rip out existing, replace with new.

If Al Viro can rewrite the guts of the VFS without hardly anyone noticing any disturbance, and the configuration system can't be done this way, something is amiss.

Not necessarily. If the gradual change infers intermediate functions that are not needed for a hard cut-over, it's no good to do all the extra work if the already-existing result is known to work. I'd certainly not drop CML1 from 2.4, but 2.5 is the time for changes.

You don't take an airplane to drive alongs autobahns for a gradual change either.

It seems to me as though the "rip out existing" issue was rather a "don't let us maintain two parts of the same type at the same time" convenience feature than a trait of CML2, and if so, keep both for some releases and drop CML1 in, say, 2.5.8.

#84 lore raw
In reply to Alexander Viro

again, if you all would spend 1/10th of your time helping as opposed to just slaming it, I'd be impressed. CML2 makes a WORLD of difference for me personally. But hey, what do I know? I'm just a network administrator, I'm not a kernel wizard, so feel free to just totally disregard my opinions...

----- Original Message -----
From: "Larry McVoy" <[email protected]>
To: "William Scott Lockwood III" <[email protected]>
Cc: "Alexander Viro" <[email protected]>; "Eric S. Raymond" <[email protected]>; <[email protected]>
Sent: Friday, February 15, 2002 7:59 PM
Subject: Re: Disgusted with kbuild developers

Quoted 1 line

On Fri, Feb 15, 2002 at 07:14:17PM -0600, William Scott Lockwood III

Quoted 1 line · wrote:
Quoted 1 line

Here's an idea Al: Put up, or shut up. Where's your code to replace

CML1

Quoted 1 line
Quoted 1 line

with? I like Eric's system, and find it's MUCH more useful in a

production

Quoted 9 lines
Quoted 1 line

environment than CML1 was. Where's your replacement idea?

I think Al's work on the kernel speaks for itself. He's been putting up just fine, thank you. And I think you're looking in the wrong place, Al has expressed no interest (that I'm aware of) in rewriting this code. As a user of the new system, a pretty typical user, his opinion counts.

I don't think that you're really getting the point. Nobody is saying that CML1 is the greatest thing since sliced bread. What they are saying is

that

Quoted 1 line

it seems to work pretty well, yes it could be better, but CML2 isn't

shaping

Quoted 1 line

up to be an improvement so much as a Eric Raymond Language exercise.

Noone

Quoted 1 line

begrudges Eric his right to come up with as many little languages as he

wants.

Quoted 1 line

But when he asks the kernel developers to use them, he'd better be

prepared

Quoted 1 line

to hear each and every thing they find wanting, and address the majority

of

Quoted 5 lines

those issues. That hasn't happened. Instead, there have been a lot of flame wars, politics, protests about Linus, etc.
--
---
Larry McVoy lm at bitmover.com

http://www.bitmover.com/lm

Quoted 1 line
#83 lore raw

Also, it apears I am no longer getting messages from the list - thanks, whom ever it was that killed me off the list. So nice of you to do that. ----- Original Message -----
From: "John Jasen" <[email protected]>
To: "Larry McVoy" <[email protected]>
Cc: "William Scott Lockwood III" <[email protected]>; "Alexander Viro" <[email protected]>; "Eric S. Raymond" <[email protected]>; <[email protected]>
Sent: Friday, February 15, 2002 10:17 PM
Subject: Re: Disgusted with kbuild developers

Quoted 9 lines

If Microsoft really wanted to kill open source, they'd put you all in the same room together with weapons and tequila.

--
-- John E. Jasen ([email protected])
-- In theory, theory and practise are the same. In practise, they aren't.

#48 lore raw
In reply to Alan Cox
Quoted 4 lines · Alan Cox

Since the information is there in CML1 to generate the list of constraints for any given option, its a reasonable assertion that the entire CML2 language rewrite is self indulgence from a self confessed language invention freak.

Correct me if I'm wrong, but there are express two different types of situations, and CML1 isn't sufficient to express the second:

1) CONFIG_FOO_OPTION requires CONFIG_FOO

2) CONFIG_SUBSYS2 requires CONFIG_SUBSYS1

The reason why #2 is different, is the desired prompting and symbol behavior for the end user.

If CONFIG_SUBSYS1=m or "", and CONFIG_SUBSYS2=y or m, then we gotta change the value of CONFIG_SUBSYS1 and options underneath CONFIG_SUBSYS1. Re-prompt for CONFIG_SUBSYS1, perhaps? If CONFIG_SUBSYS1=y, value of CONFIG_SUBSYS2 isn't affected If CONFIG_SUBSYS1="" and CONFIG_SUBSYS2="", then we gotta prompt for CONFIG_SUBSYS1, but -after- CONFIG_SUBSYS2 is prompted for.

I was tempted to introduce a "requires" token to express dependencies between subsystems, because I feel they are different from the other dependencies present,

Jeff

Signature · 3 lines
Jeff Garzik      | "I went through my candy like hot oatmeal
Building 1024    |  through an internally-buttered weasel."
MandrakeSoft     |             - goats.com
#100 lore raw
In reply to Eric S. Raymond
Quoted 9 lines · That Linux Guy

The amount of whining I've had to read today on this topic just astonishes me. People, please - devote 1/10th of the energy to helping esr finish and fix CML2 to meet Linus' expectations. Face the fact: CML1 was good for it's time, but that time is now gone.

If you spent as much time helping to get CML2 up and going as some of you have trying to kill it (while not offering anything constructive to replace it) it would have been DONE 6 months ago.

I consider CML2 an albatross now, and the maintainer does not listen to feedback.

Contrary to what you think, I'm not trying to kill CML2. In it's present state, according to kernel developers other than me, it is NOT ready for inclusion. Now, by all appearances, Eric is trying to ram CML2 down Linus's throat without resolving the problems kernel developers have with it.

I would be happy as a clam if Eric wanted to resolve the problems -we- have with CML1 and the config system, not just the problems he has with it.

Jeff

Signature · 3 lines
Jeff Garzik      | "I went through my candy like hot oatmeal
Building 1024    |  through an internally-buttered weasel."
MandrakeSoft     |             - goats.com
#101 lore raw
In reply to Eric S. Raymond
Quoted 4 lines · Dave Jones

Increased functionality I don't have a problem with, as long as other more important things are addressed. And for that matter, Linus has said to Eric "I don't care, take this out of the kernel completely leaving just oldconfig'.

That's a good point, and one I would be happy with. (or, ditch -all- current config code, and replace with the existing mconfig)

Eric's configurator definitely seems to have a place with users. Making kernel configuration easier for the masses is more than fine with me... Impacting kernel developers' productivity and workflow because of this is more of what I object to...

Jeff

Signature · 3 lines
Jeff Garzik      | "I went through my candy like hot oatmeal
Building 1024    |  through an internally-buttered weasel."
MandrakeSoft     |             - goats.com
#102 lore raw
In reply to Eric S. Raymond
Quoted 8 lines · Eric S. Raymond

Dave Jones <[email protected]>:

Quoted 6 lines

As you obviously completely ignored my previous point, I'll reiterate it.

1. You run Linus' split-into-config.in script on a 2.4 tree.
2. You add whatever Config.ins have been updated to the 2.4 config.ins
3. You regenerate with the find example I showed in the other mail.

Quoted 6 lines
Quoted 2 lines

There. 2.4 Configure.help up to date with latest symbols, but containing none of the 2.5 ones.

And you've completely ignored the real problem...which is when I get a text change for one tree, *how do I automatically propagate it to the other*? How do I *tell* that it ought to be propagated?

It's easy as hell to propagate a Config[ure].help change across trees. Linus even gave you the code to do it, when he split up Configure.help.

Configure.help is nothing but a database, with two different but easy-to-parse formats in 2.4 and 2.5.

Quoted 2 lines

Solutions that involve me doing an arbitrary and increasing amount of hand-hacking on every release are right out.

If you are hand-hacking Config[ure].help changes, you are wasting a lot of time...

Jeff

Signature · 3 lines
Jeff Garzik      | "I went through my candy like hot oatmeal
Building 1024    |  through an internally-buttered weasel."
MandrakeSoft     |             - goats.com
#103 lore raw
In reply to Eric S. Raymond

Rob Landley <[email protected]>:

Quoted 6 lines · Eric S. Raymond
Quoted 3 lines

Solutions that involve me doing an arbitrary and increasing amount of hand-hacking on every release are right out.

Um, Eric? Isn't that what being a maintainer basically means?

The problem isn't that I'm not willing to work hard. I am. It's that the level of handhacking has to be controlled below the threshold that makes it impossible.

Quoted 5 lines

Back up a bit. What would be the most minimal, stripped-down version of CML2 you could write? No eye candy, no complications, no autoconfigurator, no tree view, no frozen symbols. Just solving the core problem of configuring 2.5 in a more flexible and less buggy way than CML1, with the three interfaces (oldconfig, menuconfig, xconfig) we've got now.

The big problem isn't the code transition. It's the rulebase transition.

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#109 lore raw
In reply to Eric S. Raymond
Quoted 4 lines · Eric S. Raymond

Dave Jones <[email protected]>:

Quoted 1 line

find . -name Config.help -exec cat {} >Configure.help \;

Back-seat drivers. Answers for everything, understanding of nothing.

Did you read this in _How to Win Friends And Influence People_?

Quoted 7 lines

Sorry, it's not *nearly* that simple. For one thing, one of the fourteen billion requirements I've had dropped on me is that Marcelo doesn't want any symbols in his help file that aren't actually in the 2.4 rulebase.

So I'd have to generate a custom version for him. I used to have the machinery to do that by parsing magic comments in my master file. Until Linus blew it apart.

Linus also gave you C code which, slightly modified, could discern which CONFIG_xxx help texts were needed for a given tree.

How hard is it to maintain a database where the key for each entry is easy an unambigious?

Jeff

Signature · 4 lines
Jeff Garzik      | "Why is it that attractive girls like you
Building 1024    |  always seem to have a boyfriend?"
MandrakeSoft     | "Because I'm a nympho that owns a brewery?"
                 |             - BBC TV show "Coupling"
#110 lore raw
In reply to Eric S. Raymond
Quoted 6 lines · Eric S. Raymond

The first thing I heard, from mec two years ago, was "the CML1 code base is not salvageable". This was then and is now the unanimous opinion of the kbuild team. Not just mine; in fact, they concluded it before I entered the picture.

I have seen nothing since to make me change that opinion.

So Alan Cox's opinion counts as nothing?
mconfig counts as nothing?

Anyway, I think you are missing the point. Sure, CML1 has problems, but is your solution the one we want for the kernel? I do not think that is clear yet. And bragging about a system's use in production tells us little about its suitability for most kernel developers, or whether you've addressed the global dependencies problem, or the syntax problem, etc.

I'm sure CML2 configurator is whiz-bang, but it needs to do basic stuff like having "make oldconfig" work exactly like it does in the current config system.

Jeff

Signature · 4 lines
Jeff Garzik      | "Why is it that attractive girls like you
Building 1024    |  always seem to have a boyfriend?"
MandrakeSoft     | "Because I'm a nympho that owns a brewery?"
                 |             - BBC TV show "Coupling"
#104 lore raw
In reply to Eric S. Raymond
Quoted 2 lines · [email protected] said:

The big problem isn't the code transition. It's the rulebase
transition.

Only because you make it so.

I have no objection to the CML2 language and tools.

I've never done much more than glance at the CML1 mechanism, and as long as "$EDITOR .config && make oldconfig" continues to work as expected I doubt I'd ever pay any more attention to CML2.

I am aware that CML2 should allow me to more accurately represent some things that CML1 couldn't, and I would welcome that.

However - the thing to which I and many others object most strongly is the rulebase policy changes which appear to be inseparable from the change in mechanism. That is; we've tried to get you to separate them, and failed.

As you keep saying, CML1 is a very limited language. If CML2 cannot even represent the _existing_ CML1 ruleset, then I think you need to go back to the drawing board.

I, for one, will have no objection to a merge of CML2 with an automated translation of the CML1 rules. You can 'improve' the rulebase later, at which point each change you want to make can be given individual attention.

You've said before that an automated translation is impossible. That's not our problem - it's yours. Fix the CML2 language so it _is_ possible. If that means adding ugly hacks to make it work, so be it - do it in the expectation that you can fix the rulebase later and remove them again.

Eric, you have demonstrated repeatedly that your taste w.r.t the configuration rules is extremely, erm, 'different' from that of the majority of developers. I find it difficult to believe that a self-proclaimed 'hacker of social systems' cannot comprehend the need to strictly separate the mechanism changes from the controversial rulebase changes.

--
dwmw2

#51 lore raw
In reply to Alan Cox
Quoted 11 lines · Jeff Garzik
Quoted 4 lines · Alan Cox

Since the information is there in CML1 to generate the list of constraints for any given option, its a reasonable assertion that the entire CML2 language rewrite is self indulgence from a self confessed language invention freak.

Correct me if I'm wrong, but there are express two different types of situations, and CML1 isn't sufficient to express the second:

1) CONFIG_FOO_OPTION requires CONFIG_FOO

Quoted 8 lines

2) CONFIG_SUBSYS2 requires CONFIG_SUBSYS1

The reason why #2 is different, is the desired prompting and symbol behavior for the end user.

If CONFIG_SUBSYS1=m or "", and CONFIG_SUBSYS2=y or m, then we gotta change the value of CONFIG_SUBSYS1 and options underneath CONFIG_SUBSYS1. Re-prompt for CONFIG_SUBSYS1, perhaps?

IMHO that is a issue with the current *tools*, not with the CML1 *language*. The information about the dependences is there, a more clever tool than "make config" can use them to present a better UI.

I have a 5-year-old perl script for kernel configuration, maybe I should try to reactivate it and see ...

Gerd

Signature · 1 line
#define	ENOCLUE 125 /* userland programmer induced race condition */
#134 lore raw
In reply to Eric S. Raymond
Quoted 10 lines · Eric S. Raymond

Alexander Viro <[email protected]>:

Quoted 5 lines
Quoted 2 lines

Neither Linus nor anyone else ever said to me "Eric, it's your job to make that change." When I asked for guidance, Linus blackholed my mail.

Oh! Come and see the violence inherent in the system! Help! Help! I'm being repressed!

Your reply was pretty funny. Well, I laughed anyway.

You didn't reach for your usual potence substitute called a weapon? Didn't you?

#111 lore raw
In reply to Eric S. Raymond
Quoted 5 lines · That Linux Guy

The amount of whining I've had to read today on this topic just astonishes me. People, please - devote 1/10th of the energy to helping esr finish and fix CML2 to meet Linus' expectations. Face the fact: CML1 was good for it's time, but that time is now gone.

The fact is CML1 works fine enough for me. In fact fine engough means - I never have any problems with it. *THIS* is what makes the inclination to replace it with something else so low. And this is a good thing. No change for the sake of it is a primal principle in good software design.

#54 lore raw
In reply to Alan Cox
Quoted 10 lines · Alan Cox
Quoted 7 lines

Alan Cox <[email protected]>:

Quoted 3 lines

Since the information is there in CML1 to generate the list of constraints for any given option,

False assumption...

Do you have anything more constructive that "doesnt" to say

Quoted 9 lines
Quoted 4 lines

If you want to refute me, build it yourself. You'll get a valuable learning experience. At the end of it, I'll have earned your apology. Not that I expect to get it.

Been there, done that, got the pretty graphs. Possibly the next step is to redo the work into mconfig which already does pretty much anything we need with the existing config files parsed using a real 100% genuine yacc grammar

I second this. mconfig is a fine lean and sane design, which isn't trying to solve the "world hunger" problem.

#26 lore raw
In reply to Eric S. Raymond
Quoted 5 lines · Matthias Andree

Eric has done it, without being of kernel hacker temple's fame.

Whether he doesn't listen to other developers, I cannot tell, I got my /fetchmail/ issues resolved. Still, I share the opinion that the kernel

Fetchmail is a prime example for the fact that Eric isn't a capable software designer. Yes it is usefull, and works, but the configuration doesn't solve the common problems for which fetchmail is used in an easy way... I wouldn't like to have something like fetchmail for the kernel configuration.

#112 lore raw
In reply to Eric S. Raymond
Quoted 8 lines · Jeff Garzik
Quoted 4 lines · Dave Jones

Increased functionality I don't have a problem with, as long as other more important things are addressed. And for that matter, Linus has said to Eric "I don't care, take this out of the kernel completely leaving just oldconfig'.

That's a good point, and one I would be happy with. (or, ditch -all- current config code, and replace with the existing mconfig)

If so, then fairness of mconfig vs. CML2 tools should command that mconfig's "old" mode works properly first. I reported to Christoph that it always reprompts me about the kernel core format (a.out vs. ELF), which is something "make oldconfig" does not do with the same config, he acknowledged the bug, but I have yet to see the fix.

Quoted 4 lines

Eric's configurator definitely seems to have a place with users. Making kernel configuration easier for the masses is more than fine with me... Impacting kernel developers' productivity and workflow because of this is more of what I object to...

Does that really happen? It looks as though the conversion of the current config.in stuff has been completed as part of the CML2 suite.

#49 lore raw
In reply to Jeff Garzik
Quoted 5 lines

1) CONFIG_FOO_OPTION requires CONFIG_FOO
2) CONFIG_SUBSYS2 requires CONFIG_SUBSYS1

The reason why #2 is different, is the desired prompting and symbol behavior for the end user.

You can tell a subsystem from a module quite reliably because it has a subtree of dependant questions. Now you might get it wrong, but its still correct to prompt for the subtree of questions again since if you've just forced on eepro100 for example, you do want to be asked mmio/pio and the like

Quoted 2 lines

If CONFIG_SUBSYS1="" and CONFIG_SUBSYS2="", then we gotta prompt for CONFIG_SUBSYS1, but -after- CONFIG_SUBSYS2 is prompted for.

The graph tells you that. The only interesting case I could find is the negation one - some rules are A conflicts with B which makes the UI side much more fun

#40 lore raw
In reply to Eric S. Raymond
Quoted 7 lines · Martin Dalecki
Quoted 5 lines · Matthias Andree

Eric has done it, without being of kernel hacker temple's fame.

Whether he doesn't listen to other developers, I cannot tell, I got my /fetchmail/ issues resolved. Still, I share the opinion that the kernel

Quoted 7 lines

Fetchmail is a prime example for the fact that Eric isn't a capable software designer.
Yes it is usefull, and works, but the configuration doesn't solve the common problems
for which fetchmail is used in an easy way... I wouldn't like to have something like
fetchmail for the kernel configuration.

The point is not to discuss fetchmail ease of use or design or whatever. CML2 is much younger, and in a much more maintainable language (or so I believe, at last; Python vs. C).

The point I'm raising is that I cannot believe Eric would not care for change requests right now, that contradicts my experience.

Signature · 4 lines
Matthias Andree

"They that can give up essential liberty to obtain a little temporary
safety deserve neither liberty nor safety."         Benjamin Franklin
#105 lore raw
In reply to Eric S. Raymond

David Woodhouse <[email protected]>:

Quoted 3 lines

However - the thing to which I and many others object most strongly is the rulebase policy changes which appear to be inseparable from the change in mechanism. That is; we've tried to get you to separate them, and failed.

Failed? Hardly.

The only rulebase policy change Tom Rini was able to identify in a recent review was the magic behavior of EXPERT with respect to entries without help. Which I then removed by commenting out a single declaration.

There is a widespread myth that the CML2 rulebase is lousy with "policy changes". I don't know how it got started, but it needs to die now.

Maybe I need to write a CML2 FAQ.

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#50 lore raw
In reply to Jeff Garzik

Alan Cox <[email protected]>:

Quoted 3 lines

The graph tells you that. The only interesting case I could find is the negation one - some rules are A conflicts with B which makes the UI side much more fun

That's right. This is a CML2 require/prohibit construct. CML1 cannot express this, and it's essential for side-effect forcing to work. Jeff's observation about being tempted to introduce a `require' turns out actually to be equivalent once you see how both problems generalize.

You can't deduce these constraints from graph analysis, because they're not implicit in the if/then tree structure that is the only thing CML1 knows about.

Jeff and Alan have now almost caught up to where I was two years ago when I realized the CMl1 formalism was inadequate.

This is going in the FAQ.

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#106 lore raw
In reply to Eric S. Raymond
Quoted 7 lines · [email protected] said:

David Woodhouse <[email protected]>:

Quoted 3 lines

However - the thing to which I and many others object most strongly is the rulebase policy changes which appear to be inseparable from the change in mechanism. That is; we've tried to get you to separate them, and failed.

Failed? Hardly.

Quoted 8 lines

The only rulebase policy change Tom Rini was able to identify in a recent review was the magic behavior of EXPERT with respect to entries without help. Which I then removed by commenting out a single declaration.

There is a widespread myth that the CML2 rulebase is lousy with "policy changes". I don't know how it got started, but it needs to die now.

Each time I've even glanced at it, I've seen bogus changes. I haven't looked at it recently, so cannot refute your statement that they are now all gone. You'll forgive me if I take your assertion with a pinch of salt, though?

A good way to kill this myth, if myth it is, would be to set up a test suite, as I suggested before. You already have a 'randomconfig' for CML2, I believe? I think there's also one for CML1.

Repeatedly make a random config (for a random architecture), with either CML1 or CML2. Make oldconfig with the other CML, then with the first again. If there are any differences between the original randomconfig output and the output after the two 'oldconfig' stages, you've hit something that may be a problem.

Every time you hit such a difference, either fix it or document it and justify it. Ensure that the list of such justifications required is small, in order to improve the chance of CML2 being accepted.

You are permitted to fix these differences by modifications to the CML1 rules. The only cases where you should accept differences are where the CML1 behaviour does not accurately represent the intention of the developer, the developer has agreed with you, _and_ CML2 cannot implement the same buggy behaviour.

Note the third condition there. If CML2 _can_ implement the buggy CML1 behaviour, do so -- you can fix the rules _later_.

--
dwmw2

#113 lore raw
In reply to Eric S. Raymond

Jeff Garzik <[email protected]>:

Quoted 2 lines

I consider CML2 an albatross now, and the maintainer does not listen to feedback.

New FAQ entry:

* Eric doesn't listen to feedback.

Bill Stearns observed in February 2002 that "Eric has been very good about folding in changes". In response to feedback from the kernel mailing list, Eric has:

- introduced the prefix declaration so that symbol names can be expressed with or without the CONFIG_

- added a recovery mechanism so the configurator can do someting with inconsistent configs (this one was Alan Cox's issue).

- removed the magic tie declaration that made symbols without help only visible in EXPERT mode

- reorganized the rulebase to minimize stuff in the root rules file

- added the capability to annotate constraint violations with text messages in English or the language of your choice.

- sped up the rulebase compiler by a factor of six

and many other changes comprising months of work.

Quoted 3 lines

I would be happy as a clam if Eric wanted to resolve the problems -we- have with CML1 and the config system, not just the problems he has with it.

What problems do you have with CML1? Name them and I will make sure they are not issues in CML2.

I've been begging for feedback for many months. Pleases *get specific*. Instead of muttering that Eric refuses to listen, *give me something to listen to*!

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#127 lore raw
In reply to Eric S. Raymond

Jeff Garzik <[email protected]>:

Quoted 2 lines

Impacting kernel developers' productivity and workflow because of this is more of what I object to...

I still don't get why you think CML2 is going to interfere with your workflow. Would you explain this to me, please?

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#128 lore raw
In reply to Eric S. Raymond

Jeff Garzik <[email protected]>:

Quoted 2 lines

How hard is it to maintain a database where the key for each entry is easy an unambigious?

It's not that simple, either. You have to track the symbols that are *supposed* to be different between trees. The devil is in the details -- and it's a big devil.

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#108 lore raw
In reply to Eric S. Raymond

----- Original Message -----
From: "Eric S. Raymond" <[email protected]>
To: "David Woodhouse" <[email protected]>
Cc: "Rob Landley" <[email protected]>; "Dave Jones" <[email protected]>; "Larry McVoy" <[email protected]>; "Arjan van de Ven" <[email protected]>; <[email protected]>
Sent: Saturday, February 16, 2002 8:57 AM
Subject: Re: Disgusted with kbuild developers

Quoted 5 lines

David Woodhouse <[email protected]>:

Quoted 3 lines

However - the thing to which I and many others object most strongly is the rulebase policy changes which appear to be inseparable from the change in mechanism. That is; we've tried to get you to separate them, and failed.

Quoted 8 lines

Failed? Hardly.

The only rulebase policy change Tom Rini was able to identify in a recent review was the magic behavior of EXPERT with respect to entries without help. Which I then removed by commenting out a single declaration.

There is a widespread myth that the CML2 rulebase is lousy with "policy changes". I don't know how it got started, but it needs to die now.

The last time I played with CML2 (~2 months ago) I didn't even make it through a full 'make config' before I had to stop on account of nausea. It was seemingly overrun with gratuitous changes that were (to me, anyway) useless and annoying.

Examples:
* It started asking me vague questions like how old my computer was in order to draw some magic conclusion about my configuration needs. *I* know my hardware. Don't waste my time with Aunt Tillie questions that will quite possibly draw the wrong conclusion for my needs.

* The 'make config' UI was subtly changed from the CML1 version. Why? I haven't seen any complaints on the list about the UI.

* The 'make menuconfig' UI was significantly changed. Again, why? Haven't seen any complaints about the old one.

Maybe these issues don't qualify as "policy change" but it seems obvious that there is some other agenda being followed here besides replacing the CML1 language with one that can accurately represent the rulebase. I'm in favor of the later, but definitely not if it comes inseparably packaged with the former.

--Adam

#130 lore raw

CML2 feedback (was Re: Disgusted with kbuild developers)

In reply to Eric S. Raymond
Quoted 6 lines · Eric S. Raymond

What problems do you have with CML1? Name them and I will make sure they are not issues in CML2.

I've been begging for feedback for many months. Pleases *get specific*. Instead of muttering that Eric refuses to listen, *give me something to listen to*!

(cc'ing Linus)

1) Remove global dependencies.

Quoted 4 lines

source "arch/um/rules.cml"
source "arch/i386/rules.cml"
source "arch/alpha/rules.cml"
source "arch/sparc/rules.cml"

[...]

All symbols should not be included on every config run. The architectures currently create their own master rules file, which then selectively includes other config files. You break this and remove control (and flexibility) by enforcing something globally which was previously done on a per-arch basis.

2) As discussed in December (as well as before and after that thread), the --eventual-- direction to go is that a single file will contain all meta information about a driver or set of drivers. Config.help entries, Config.in entries, Makefile rules, etc. The idea is to group information about a particular object in the same location. So, considering (a) that Config.help now exists and (b) this eventual direction, splitting up rules and symbols into completely separate files is not the greatest idea... when we eventually want to integrate rules and symbols.

Take a look at what we find in Donald Becker's virgin source tree, in winbond-840.c:

Quoted 11 lines

/* Automatically extracted configuration info:
probe-func: winbond840_probe
config-in: tristate 'Winbond W89c840 Ethernet support' CONFIG_WINBOND_840

c-help-name: Winbond W89c840 PCI Ethernet support
c-help-symbol: CONFIG_WINBOND_840
c-help: The winbond-840.c driver is for the Winbond W89c840 chip.
c-help: This chip is named TX9882 on the Compex RL100-ATX board.
c-help: More specific information and updates are available from
c-help: http://www.scyld.com/network/drivers.html
*/

All the information is in one location. Now, Linus said not to embed this information in the source, but put it in a metadata file. But other than that, this is a small and simple example of what the metadata config files might eventually look like.

So, my own personal opinion is that CML2 should follow these suggestions :)

Regards,

Jeff

Signature · 4 lines
Jeff Garzik      | "Why is it that attractive girls like you
Building 1024    |  always seem to have a boyfriend?"
MandrakeSoft     | "Because I'm a nympho that owns a brewery?"
                 |             - BBC TV show "Coupling"
#131 lore raw
In reply to Eric S. Raymond
Quoted 7 lines · Eric S. Raymond

Jeff Garzik <[email protected]>:

Quoted 2 lines

Impacting kernel developers' productivity and workflow because of this is more of what I object to...

I still don't get why you think CML2 is going to interfere with your workflow. Would you explain this to me, please?

(please read "CML2 feedback" message before this...)

Ideally in the future I can add and update a driver's makefile and configuration information by patching drivers/net/netdriver.c and drivers/net/netdriver.conf, and touch absolutely no other files.

Jeff

Signature · 4 lines
Jeff Garzik      | "Why is it that attractive girls like you
Building 1024    |  always seem to have a boyfriend?"
MandrakeSoft     | "Because I'm a nympho that owns a brewery?"
                 |             - BBC TV show "Coupling"
#67 lore raw
In reply to Larry McVoy
Quoted 4 lines · Matthias Andree

The point is not to discuss fetchmail ease of use or design or whatever. CML2 is much younger, and in a much more maintainable language (or so I believe, at last; Python vs. C).

I am not so convinced about this "much more" in practice. As an "interested observer", but not a Python programmer, I watch for quite a while a set of programs written in Python which affect me directly; namely Red Hat configuration/installation tools. Literally for years and many releases bombs and inscrutable Python tracebacks were the constant element of this picture. These tools, after an obvious and long effort, are now finally in much better shape then they used to be but bugs are regularly reintroduced with every change.

I do not think that this happens because Red Hat has lazy or incompetent people but because these problems are hard and Python is quite far from silver bullet and "sliced bread" its proponent wants us to believe. I rather suspect that all behind the scenes machinations by Python rather mask difficulties and make harder to eradicte those bugs.

I also cannot help not to notice that the previous bit flare-up about CML2 on lkml was quelled to a great extent when somebody annouced that he is rewriting required tools in C. I had an impression that most people then shrugged "Ok, so Eric will prototope in whatever he feels comfortable with, we will have something acceptable later and we will see how this works". Now it turns out the the project got abandoned so a requirement for a huge blob of a language (as opposed to Python 1.5 which is quite smaller), which most developers do not have or need for anything else, is still there. Hm..., smells very backdoor even if it was not intended that way.

Michal

#114 lore raw
In reply to Eric S. Raymond
Quoted 11 lines · Eric S. Raymond

Jeff Garzik <[email protected]>:

Quoted 2 lines

I consider CML2 an albatross now, and the maintainer does not listen to feedback.

New FAQ entry:

* Eric doesn't listen to feedback.

Bill Stearns observed in February 2002 that "Eric has been very good about folding in changes". In response to feedback from the kernel mailing list, Eric has:

[...]

Eric has selectively listened to feedback he found interesting and still persist in resisting most others.

You ignored mine so far, even if they are like the ones made by many others.

Make the whole thing ___***IDENTICAL***___ to CML1. Do a formal translation of CML1 into CML2.

Show us that you are clever enough to do so, even if it's not particularly interesting and challenging to you.

Show us that you can listen to this simple feedback.

Acknoledge that the feedback went through.

Don't tell us that's not doable. Do it and show us that you can do a perfect translation of CML1 into CML2 with all CML1 structural flaws.

Submit that, and only that.

Do you copy? Please acknoledge that you listened to this very feedback.

Nicolas

#16 lore raw

Possible breakthrough in the CML2 logjam?

In reply to Arjan van de Ven

Jeff Garzik <[email protected]>:

Quoted 3 lines

Ideally in the future I can add and update a driver's makefile and configuration information by patching drivers/net/netdriver.c and drivers/net/netdriver.conf, and touch absolutely no other files.

That's very much the direction I'd like to go in, Jeff. I'm surprised and delighted to hear you say this. You're actually anticipating my plans for six months down the road. Maybe we have some common ground here.

One of the objectives of the CML2 language design is to make it amenable to being generated by a metadata analyzer. I mean some very specific things by this.

One important property which CML2 declarations have that CML1 syntax does not is that (a) they're not order-dependent, and (b) they have strong locality (that is, the syntactic context of a single declaration holds all the semantic information -- you don't have to go monkey-climbing up a bunch of enclosing syntax to parse it, or *generate* a bunch of enclosing syntax to express it).

These properties tremendously simplify generating a rulebase from (so far hypothetical) analysis tools. My first step would be to automatically generate CML2 bus-guard information from annotations in the driver sources. Once I write a tool that can do that, I can whack about 25% of the rulebase, including most of the parts that are a maintainance headache.

My longer-term plan is to whittle away at the manually-maintained rulebase until nothing but menu structure and a handful of cross- directory requirements are left. Everything else should be generated by a program that turns source-code metadata into stereotyped CML2 markup.

Even a lot of the menu structure might be generatable, requiring it to be specified only in exception cases where as a matter of UI design choice you don't want to track the code hierarchy.

This has been part of my long-term plan since about eighteen months ago. It's had a major, *major* impact on the language design. In particular it's one of the reasons visibility and implication can be declared separately from the menu structure.

If you go back and look at the language design from this point of view, I think many things you might not have seen the point of before will become clearer.

Note well two points:

1. This can't practicably be done in CML1. CML1 markup has crappy locality; the metadata analyzer would have to carry around way too much state about other symbols in order to generate the markup for any given one.

2. This design basically demands a single-apex tree. Otherwise I don't think you can get the consistency-checking right -- I haven't been able to invent a method to do it, anyway.

So if you want this, please start backing CML2 and contributing in a positive way. I know how to get where you want to go. CML2 is specifically intentionally designed to make it possible, and I have the will to follow through.

But for these good things to happen, CML2 *got to go in*. I cannot both continue the enormous effort of maintaining a parallel rulebase and move the ball forward towards automatic rule generation from metadata and other good things. That's what I want to be working on.

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#93 lore raw
In reply to Eric S. Raymond

Alan Cox <[email protected]>:

Quoted 5 lines
Quoted 3 lines

The kernel rulebase cannot. The real issue here is whether the CML1 language carries sufficient information to support things like side-effect deduction. And the answer is "no".

Thats an opinion not an answer. What doesn't it contain ?

New FAQ entry:

* Inventing CML2 was unnecessary, since CML1 carries enough information to do consistency checking and side-effect forcing.

Jeff Garzik observed:

Quoted 3 lines

I was tempted to introduce a "requires" token to express dependencies between subsystems, because I feel they are different from the other dependencies present,

Alan followed up with:

Quoted 2 lines

The only interesting case I could find is the negation one - some rules are A conflicts with B which makes the UI side much more fun

Jeff and Alan have put their finger neatly on one of the key bits CML2 can do that CML1 cannot -- express cross-directory dependencies in such a way that the configurator can force side effects in both directions. This is, in fact, the very rock on which my original attempt to save CML1 foundered after six weeks of effort.

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#107 lore raw
In reply to Eric S. Raymond

David Woodhouse <[email protected]>:

Quoted 3 lines

A good way to kill this myth, if myth it is, would be to set up a test suite, as I suggested before. You already have a 'randomconfig' for CML2, I believe? I think there's also one for CML1.

There is no randomconfig for CML2.

Quoted 9 lines

Repeatedly make a random config (for a random architecture), with either CML1 or CML2. Make oldconfig with the other CML, then with the first again. If there are any differences between the original randomconfig output and the output after the two 'oldconfig' stages, you've hit something that may be a problem.

Every time you hit such a difference, either fix it or document it and justify it. Ensure that the list of such justifications required is small, in order to improve the chance of CML2 being accepted.

This cycle is what I've been going through with a lot of my beta testers.

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#94 lore raw
In reply to Eric S. Raymond
Quoted 5 lines

Jeff and Alan have put their finger neatly on one of the key bits CML2 can do that CML1 cannot -- express cross-directory dependencies in such a way that the configurator can force side effects in both directions. This is, in fact, the very rock on which my original attempt to save CML1 foundered after six weeks of effort.

You can force a side effect in both directions. The language provides the information to do that, the current -toolset- can't handle this.

At any point you ask a question you can "wind back" and compute the set of changes that are needed and re-ask only the needed questions.

#21 lore raw
In reply to Eric S. Raymond

Nicolas Pitre <[email protected]>:

Quoted 2 lines

Show us that you're able to write a 1 for 1 functional correspondance between CML1 and CML2 and propose that for inclusion into 2.5.

This requirement is absurd. When someone designs a new VM, we don't demand that it crash or lock up the system in exactly the same way that the old one did before it can go into the kernel.

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#17 lore raw

Re: Possible breakthrough in the CML2 logjam?

In reply to Arjan van de Ven
Quoted 4 lines · Eric S. Raymond

But for these good things to happen, CML2 *got to go in*. I cannot both continue the enormous effort of maintaining a parallel rulebase and move the ball forward towards automatic rule generation from metadata and other good things. That's what I want to be working on.

Global dependencies... CML1 doesn't have this now, and it needs never to have it. This is no point in merging a design change of that magnitude only to take it away later on. Further, merging a rulebase which contains such dependencies would be a huge mistake that might take years to undo. drivers/net/rules.cml doesn't need S/390 stuff in it, AFAICT, and that is a simple example of a bug found in many of the rules.cml files.

Jeff

Signature · 4 lines
Jeff Garzik      | "Why is it that attractive girls like you
Building 1024    |  always seem to have a boyfriend?"
MandrakeSoft     | "Because I'm a nympho that owns a brewery?"
                 |             - BBC TV show "Coupling"
#117 lore raw
In reply to Eric S. Raymond

Nicolas Pitre <[email protected]>:

Quoted 8 lines

Make the whole thing ___***IDENTICAL***___ to CML1. Do a formal translation of CML1 into CML2.

Show us that you are clever enough to do so, even if it's not particularly interesting and challenging to you.

Show us that you can listen to this simple feedback.

Quoted 8 lines

Acknoledge that the feedback went through.

Don't tell us that's not doable. Do it and show us that you can do a perfect translation of CML1 into CML2 with all CML1 structural flaws.

Submit that, and only that.

Do you copy? Please acknoledge that you listened to this very feedback.

I listened.

Would you ask someone designing a new VM to make it crash and hang exactly the same way the old one did?

Do you demand that a rewrite of a disk driver have the same data-corruption bugs as the original before it can go into the tree, and tell the developer to add fixes later?

Pragmatically, the point of rewriting a system is to *fix bugs*.

Let's suppose we ignored this point for a moment. Let's also suppose that what you were demanding were not rendered horribly painful and perhaps impossible by the difference between CML1's imperative style and CML2's declarative one.

How the hell do you possibly think I could possibly stay motivated under that constraint? Nobody is paying me to do this. I'm a volunteer; I need to produce good art, not waste time slavishly recreating old errors just because a few people are unreasonably fearful of change.

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#124 lore raw
In reply to Eric S. Raymond
Quoted 9 lines · Eric S. Raymond

Let's suppose we ignored this point for a moment. Let's also suppose that what you were demanding were not rendered horribly painful and perhaps impossible by the difference between CML1's imperative style and CML2's declarative one.

How the hell do you possibly think I could possibly stay motivated under that constraint? Nobody is paying me to do this. I'm a volunteer; I need to produce good art, not waste time slavishly recreating old errors just because a few people are unreasonably fearful of change.

If you are not prepared to evolve the system, you are not familiar with kernel development... Prove that CML2 is good and needed by breaking it up into steps, reaching the final goal... Good ole Al Viro's patch progressions are an excellent example to emulate :)

Jeff

Signature · 4 lines
Jeff Garzik      | "Why is it that attractive girls like you
Building 1024    |  always seem to have a boyfriend?"
MandrakeSoft     | "Because I'm a nympho that owns a brewery?"
                 |             - BBC TV show "Coupling"
#129 lore raw
In reply to Eric S. Raymond
Quoted 3 lines

It's not that simple, either. You have to track the symbols that are *supposed* to be different between trees. The devil is in the details -- and it's a big devil.

(Not tested but something like this ought to split them neatly out of
a master tree into the 2.4 format)

for i in $(find . -name "[Cc]onfig.in" -print)
do
grep CONFIG_ '$i'
done | sort -u |
while read x
do

	grep -B1 -A200 '$x' MasterConfigHelpFile |
	( 
		read A
		read B
		cat > frob
	)
	ed frob << 'FOO'
	/^$
	.,$d
	wq
	'FOO' >/dev/null
	cat frob

done

#96 lore raw
In reply to Eric S. Raymond

Alan Cox <[email protected]>:

Quoted 5 lines

You can force a side effect in both directions. The language provides the information to do that, the current -toolset- can't handle this.

At any point you ask a question you can "wind back" and compute the set of changes that are needed and re-ask only the needed questions.

I spent over a month in early 2000 trying a similar approach. I tried it with CML1, and I tried it with increasingly enriched dialects of CML1 (magic comments carrying extra semantic information, that sort of thing).

The results were (a) ugly, and (b) broken. I struggled against this for a long time, because I knew what a horrible revolving bitch and maintaining a parallel rulebase in a new formalism was going to be.

As you no doubt realize, the problem of deducing the forcing information from CMl1 markup is efectively equivalent to the problem of writing a mechanical CML1-to-CML2 translator. So I have a
suggestion: if you want to prove that it's possible to extract all the info for side-effect forcing from CML1, do it by writing such a translator.

I believe you will fail, as I did and as Jeff Garzik implicitly predicted. If you fail, the process will teach you what I had to learn the hard way two years back. If you succeed, people who are whingeing about wanting a bug-for-bug rulebase translation will get what they want.

Don't tell me to do it. Been there, done that, have the battle scars. If there were any way I could have avoided maintaining my own rulebase, you better believe I'd have done it.

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#95 lore raw
In reply to Alan Cox
Quoted 8 lines · Alan Cox
Quoted 6 lines

Jeff and Alan have put their finger neatly on one of the key bits CML2 can do that CML1 cannot -- express cross-directory dependencies in such a way that the configurator can force side effects in both directions. This is, in fact, the very rock on which my original attempt to save CML1 foundered after six weeks of effort.

You can force a side effect in both directions.

Yup, good point.

I mentioned "requires" as a way of avoiding 'both directions' in some cases.

Jeff

Signature · 4 lines
Jeff Garzik      | "Why is it that attractive girls like you
Building 1024    |  always seem to have a boyfriend?"
MandrakeSoft     | "Because I'm a nympho that owns a brewery?"
                 |             - BBC TV show "Coupling"
#98 lore raw
In reply to Eric S. Raymond
Quoted 1 line · Eric S. Raymond

I believe you will fail, as I did and as Jeff Garzik implicitly predicted.

no, I didn't, don't put imaginary words in my mouth.

Signature · 4 lines
Jeff Garzik      | "Why is it that attractive girls like you
Building 1024    |  always seem to have a boyfriend?"
MandrakeSoft     | "Because I'm a nympho that owns a brewery?"
                 |             - BBC TV show "Coupling"
#22 lore raw
In reply to Eric S. Raymond
Quoted 7 lines · Eric S. Raymond

Nicolas Pitre <[email protected]>:

Quoted 2 lines

Show us that you're able to write a 1 for 1 functional correspondance between CML1 and CML2 and propose that for inclusion into 2.5.

This requirement is absurd. When someone designs a new VM, we don't demand that it crash or lock up the system in exactly the same way that the old one did before it can go into the kernel.

When someone design a new VM, and decides that it must be written in a totally new language with a new compiler, that someone will certainly face big resistance unless it can be proven that the new language can do exactly the same as the old one so other developers can get used to it first... especially if the old VM doesn't crash the system that often or maybe never for many users.

So yes, it might look absurd from your point of view but it's not for most people not familiar with CML2. And if that's what most of your opponants are asking for why don't you give them just that? Prove them that you're able to _split_ the concepts apart i.e. first the language vocabulary and tools, then the improved grammar, then the advanced configuration features, then the ultimate philosophical changes, etc.

Wake up Eric!

If people want A, B, C, agree somewhat with D, but have problems with E, do you realise that they will reject the whole thing at once if the only way you can present things is by submiting ABCDEFG indistinguishly?

Do things in steps. First the struct translation of CML1 into CML2 with the new coherent frontends. That's it. If you can't do that then give up now and don't spent more time on CML2 because it will never go anywhere as the Linux kernel is concerned.

Nicolas

#97 lore raw
In reply to Eric S. Raymond
Quoted 4 lines

of writing a mechanical CML1-to-CML2 translator. So I have a
suggestion: if you want to prove that it's possible to extract all the info for side-effect forcing from CML1, do it by writing such a translator.

I'd rather focus on making the existing tools better. If you want to convert the data into Erics pet format on the fly *you* do the work.

#36 lore raw

Re: Possible breakthrough in the CML2 logjam?

In reply to Arjan van de Ven

Jeff Garzik <[email protected]>:

Quoted 7 lines

Global dependencies... CML1 doesn't have this now, and it needs never to have it. This is no point in merging a design change of that magnitude only to take it away later on. Further, merging a rulebase which contains such dependencies would be a huge mistake that might take years to undo. drivers/net/rules.cml doesn't need S/390 stuff in it, AFAICT, and that is a simple example of a bug found in many of the rules.cml files.

All right. There are a couple of ways we can attack this -- I have some ideas. But I want a meta-question answered first. If we solve this issue, *are you on board*?

I've got to get off the fscking treadmill I'm on now where I'm spending so much time on the parallel-rulebase maintainance and the flaming politics that I can't really get anything else done.

After what you wrote upthread, I don't think you want me to be stuck either. Fact: ain't nobody else visible with both the motivation and skills to tackle what you want done. I've been thinking about metadata-centered configuration and consciously bending CML2's design towards it for longer than anyone else has even been considering the problem.

I *will* get us there -- I think the last two years have demonstrated my determination. But to do it, I need alliance rather than obstruction. I need you to tell Linus that your concerns have been met and sponsor CML2 to go in, so I can stop perpetually re-fighting old battles.

Because we agree on getting to metadata-centered configuration, your requirements list now appears to have shrunk to one. You want to "eliminate global depencies". I hear that. I got it.

What I want to know is that this is not a proxy for "CML2 can never be good enough" and that you'll be pulling more objections out of your butt till the Sun goes nova. Because if that's true there's no point in my trying to work with you.

But if "eliminate global depencies" is it, we can be allies, because ultimately we both want to get the config system to the same place. I've taken the first, biggest step -- from imperative code to declaration/constraint language. The second -- from a declaration/constraint language to a metadata-centered system -- will be easier.

A positive answer may be "Yes, that's it, let's go forward", or "Almost, there are a couple other minor and negotiable issues *which I will now list*" (where "Minor and negotiable" = "I don't have to scrap my architecture").

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#37 lore raw

Re: Possible breakthrough in the CML2 logjam?

In reply to Arjan van de Ven
Quoted 6 lines · Eric S. Raymond

But if "eliminate global depencies" is it, we can be allies, because ultimately we both want to get the config system to the same place. I've taken the first, biggest step -- from imperative code to declaration/constraint language. The second -- from a declaration/constraint language to a metadata-centered system -- will be easier.

Well, let's simmer things down a bit and see what other people have to say. Maybe I'm completely off base.

But to answer the question which the subtext seemed to asking (at least to me), no, there is no vendetta against you. And for the record on a specific detail, I have no problem with python use. If I have no major objection based purely on technical grounds, that what you'll be hearing from me.

And further, in 2.5.x series at least, minor objections can be worked through after a kernel merge. But major objections... that's not the time when something -must- -go- -in-.

Regards,

Jeff

Signature · 4 lines
Jeff Garzik      | "Why is it that attractive girls like you
Building 1024    |  always seem to have a boyfriend?"
MandrakeSoft     | "Because I'm a nympho that owns a brewery?"
                 |             - BBC TV show "Coupling"
#118 lore raw
In reply to Eric S. Raymond
Quoted 4 lines · Eric S. Raymond

Nicolas Pitre <[email protected]>:

Quoted 1 line

Do you copy? Please acknoledge that you listened to this very feedback.

I listened.

Good. We therefore can make progress.
By your comments below we now know that you listened and that it's extremely painful for you to admit the truth. But you must understand the nuances if you want to go somewhere with CML2.

Quoted 6 lines

Would you ask someone designing a new VM to make it crash and hang exactly the same way the old one did?

Do you demand that a rewrite of a disk driver have the same data-corruption bugs as the original before it can go into the tree, and tell the developer to add fixes later?

Your examples are flawed. The person who fix bugs in the VM doesn't use any other language than the original one. The person who do a rewrite of a disk driver doesn't change anything in the struct rational purpose of reading and writing blocks of bytes on a media.

Quoted 1 line

Pragmatically, the point of rewriting a system is to *fix bugs*.

Indeed! However CML2 is __way__ more than only bug fixing. If you qualify your current CML2 inclusion proposal as only "bug fixes" it's much too much under estimating the work you did and the effort you put in it.

I'm sure people will agree with the fact that the CML2 syntax is much more enjoyable to read and work with. The fact that all frontends are using the same parser eliminates bugs already. That's the part most people have nothing against. For god sake submit those parts and leave the rest for a later discussion!

Admit that in the tremendous work you did, there are parts that people will disagree with. Don't spoil it all by not separating things and only presenting it as a big unknown blat!

For people to have confidence in CML2 it must be proposed in multiple incremental changes -- changes that are small enough (either conceptually or physically) for people to be able to grok them without too much effort. You must also be prepared to see people wanting to modify things in some ways you didn't think about along the process. That's why only producing a strict CML1 --> CML2 translation would already be a big enough step for now.

Quoted 4 lines

Let's suppose we ignored this point for a moment. Let's also suppose that what you were demanding were not rendered horribly painful and perhaps impossible by the difference between CML1's imperative style and CML2's declarative one.

Do what's necessary so the translation is a near as possible and the result in say 'make config' is the same (same questions as before). You certainly can do that with some bits of awk. If you can't come with something similar, you're then already admiting yourself that CML2 is so big a change that people are right to be afraid of it.

Quoted 4 lines

How the hell do you possibly think I could possibly stay motivated under that constraint? Nobody is paying me to do this. I'm a volunteer; I need to produce good art, not waste time slavishly recreating old errors just because a few people are unreasonably fearful of change.

First, by doing what I described above, you'll please more people than you can imagine. Next you must be prepared to face the possibility that your art might not be adopted integrally. You should be motivated by the percentage that gets adopted in the end.

When Al Viro decides to rewrite the VFS, he doesn't submit it to Linus as a single big opaque patch to replace the old code at once. If he did, he would probably still be waiting for Linus to merge his art.

Note the nuance: we don't criticize your art but rather the way you present it. It might be the best configuration tool ever, but untroduce it in incremental steps and leave some slack between them for those who didn't work on it for the last year to digest them. Please be wise so not to waste your art.

Nicolas

#41 lore raw

Re: Possible breakthrough in the CML2 logjam?

In reply to Eric S. Raymond
Quoted 3 lines

I need you to tell Linus that your concerns have been met and sponsor CML2 to go in, so I can stop perpetually re-fighting old battles.

That's a fine thing for anyone and everyone to say *after* they have used the system and like it.

If you are asking for a blessing in advance, which is how I read that, I would think there is zero chance of that happening, it's not how work is done on the kernel.

Signature · 2 lines
---
Larry McVoy            	 lm at bitmover.com           http://www.bitmover.com/lm 
#62 lore raw

Re: Possible breakthrough in the CML2 logjam?

In reply to Dave Jones

Larry McVoy <[email protected]>:

Quoted 10 lines
Quoted 3 lines

I need you to tell Linus that your concerns have been met and sponsor CML2 to go in, so I can stop perpetually re-fighting old battles.

That's a fine thing for anyone and everyone to say *after* they have used the system and like it.

If you are asking for a blessing in advance, which is how I read that, I would think there is zero chance of that happening, it's not how work is done on the kernel.

We're talking about design objections here. Specific objections to actual CML2 bugs, including rulebase and UI bugs, are a different level. What I am asking is if Jeff will bless the *architecture* provided the global- dependency issue is met.

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#63 lore raw

Re: Possible breakthrough in the CML2 logjam?

In reply to Dave Jones
Quoted 8 lines · Eric S. Raymond

Larry McVoy <[email protected]>:

Quoted 7 lines
Quoted 3 lines

I need you to tell Linus that your concerns have been met and sponsor CML2 to go in, so I can stop perpetually re-fighting old battles.

That's a fine thing for anyone and everyone to say *after* they have used the system and like it.

Quoted 8 lines
Quoted 3 lines

If you are asking for a blessing in advance, which is how I read that, I would think there is zero chance of that happening, it's not how work is done on the kernel.

We're talking about design objections here. Specific objections to actual CML2 bugs, including rulebase and UI bugs, are a different level. What I am asking is if Jeff will bless the *architecture* provided the global- dependency issue is met.

Larry's right. I won't (and notice, did not in previous e-mail) provide a pre-blessing. I will promise to be fair.

But as I said, let's wait a bit and see what others say. Alan for example noted that bit about improving the existing tools.

Jeff

Signature · 4 lines
Jeff Garzik      | "Why is it that attractive girls like you
Building 1024    |  always seem to have a boyfriend?"
MandrakeSoft     | "Because I'm a nympho that owns a brewery?"
                 |             - BBC TV show "Coupling"
#59 lore raw

Re: Possible breakthrough in the CML2 logjam?

In reply to Robert Love
Quoted 8 lines · Eric S. Raymond

Larry McVoy <[email protected]>:

Quoted 7 lines
Quoted 3 lines

I need you to tell Linus that your concerns have been met and sponsor CML2 to go in, so I can stop perpetually re-fighting old battles.

That's a fine thing for anyone and everyone to say *after* they have used the system and like it.

Quoted 8 lines
Quoted 3 lines

If you are asking for a blessing in advance, which is how I read that, I would think there is zero chance of that happening, it's not how work is done on the kernel.

We're talking about design objections here. Specific objections to actual CML2 bugs, including rulebase and UI bugs, are a different level. What I am asking is if Jeff will bless the *architecture* provided the global- dependency issue is met.

See your quote above which contains "and sponsor CML2 to go in". Code is what goes in. Having the right architecture is great, we all agree, but what goes in is code. So your question above is basically "if I do this will you pressure Linus to accept my *code*". The answer to that should always be "no". You're trying to do an end run around the process. The process here is to let people see the changes, try the changes, refine the changes, and when they are ready, Linus accepts the changes. Nowhere in that process is any deal making. What you are doing is a lot like the skating judges, making deals. Your code should go in evaluated on its merits. That's the beauty of the system. It doesn't matter who *you* are, the code is the only thing. So you can be universally loved and your code might not make it in. You can be universally hated, and your code can make it in. That's a pretty cool system and I'd suggest that you stop trying to work around it and just make your code be something that people want. Then it will go in. No deals necessary.

Signature · 2 lines
---
Larry McVoy            	 lm at bitmover.com           http://www.bitmover.com/lm 
#64 lore raw

Re: Possible breakthrough in the CML2 logjam?

In reply to Dave Jones

Jeff Garzik <[email protected]>:

Quoted 2 lines

Well, let's simmer things down a bit and see what other people have to say. Maybe I'm completely off base.

Jeff, I'm not asking "other people". I'm asking *you*.

You're one of the people Linus says he trusts. Linus has said, explicitly, to myself and Dave Jones, on this very issue, "get me out of the loop". My take is that if you switch from opposing CML2 to supporting it, the political wars will probably be about over.

I hope the prospect of actually getting to a metadata-centered configuration system in our lifetime will be sufficient incentive for you to do so. Oh lovely dream...I could have a prototype metadata-to-CML2-bus-guards translator in less than two weeks, I think, if I didn't have to maintain the CML2 rulebase all by myself. I want to go there.

Quoted 5 lines

But to answer the question which the subtext seemed to asking (at least to me), no, there is no vendetta against you. And for the record on a specific detail, I have no problem with python use. If I have no major objection based purely on technical grounds, that what you'll be hearing from me.

OK. Is "global dependencies" your sole technical showstopper? If so, can we dispose of the ill-defined "CML2 will fuck up my workflow" thing?

If you tell me yes, we can move to a discussion of why global dependencies are a problem and how to solve it.

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#132 lore raw
In reply to Eric S. Raymond
Quoted 10 lines · Eric S. Raymond

Rob Landley <[email protected]>:

Quoted 5 lines · Eric S. Raymond
Quoted 2 lines

Solutions that involve me doing an arbitrary and increasing amount of hand-hacking on every release are right out.

Um, Eric? Isn't that what being a maintainer basically means?

The problem isn't that I'm not willing to work hard. I am. It's that the level of handhacking has to be controlled below the threshold that makes it impossible.

That's still not the point. You had great tools to do a job Linus -wasn't interested in having done-. Unsuprisingly, Linus broke them. Linus wasn't trying to make more work for you as 2.5 help file maintainer, he simply didn't care about side projects outside of 2.5.

It's still a question of defining the work you're expected to do. Keeping 2.4 and 2.5 in sync was a job you invented. Initially it was easy, later it stopped being easy, and eventually it became a very hard problem indeed to the point where you had to stop doing it. But the dependency between the 2.4 and 2.5 help files was always completely artificial. That you thought maintaining it was expected of you all along was an honest miscommunication, but that's water under the bridge now.

Quoted 8 lines
Quoted 6 lines

Back up a bit. What would be the most minimal, stripped-down version of CML2 you could write? No eye candy, no complications, no autoconfigurator, no tree view, no frozen symbols. Just solving the core problem of configuring 2.5 in a more flexible and less buggy way than CML1, with the three interfaces (oldconfig, menuconfig, xconfig) we've got now.

The big problem isn't the code transition. It's the rulebase transition.

Then why confuse the two?

Eric, "make menuconfig" is now green text on a black background. The old one was black and yellow text on grey boxes with a blue border. They look NOTHING alike. The tab-through buttons at the bottom have gone away, the list now takes up the whole screen horizontally and vertically instead of being confined to a curses box, and the rest of the keys you use to do stuff are different. (q does nothing in the old menuconfig, x gives you the option to abort, and cursor left and right changes which button is highlighted. Looking back, yes "?" pulls up help in CML1 but I didn't need to know that. I never used it, I always tabbed over to "help" and hit enter. And I never used "x" to exit, I tabbed over to the "exit" button and used it until I got out of the top level menu. It's all I ever needed to know to make it work...)

I personally don't care about the other configurators since the only one I ever use is menuconfig. And after using CML2 for a while, I'll even grant that the old CML1 interface was clunky and the new one is probably an improvement. But that's not the POINT.

It IS a different interface. For no apparent reason except that you don't want to implement the old one. Pull them up side by side in two different windows and try to do the same config in paralell. The difference is pretty darn obvious.

Remember when I asked for a blank line at the top of menuconfig and you went "no, that's a waste of screen space" and refused to even consider it? We went back and forth for ten minutes on this issue. Look at how MUCH of the old menuconfig is wasted in drawing boxes and drop shadows: there's only ten lines of actual menu. (Your code doesn't have any blank lines in it either. Mine does. Does this sound like a coding style flame war to you? Do you see how it's an issue on which a judgement call from you might not suddenly resolve all debate?)

Your aesthetic judgement of how the interface should look is not the point. Can you or can you not make one that looks and acts like the old one? Yes or no? (Yes it IS a trivial problem relative to the rest of the codebase, but it's one that hasn't been addressed. And considering the UI is what people actually see, and it's different, you're giving a horrible first impression on the compatability front.)

I don't care if it's "better". New Coke was "better". It won all the blind taste tests hands down. The fact CML2 asks the same questions and produces exactly the same .config files as the old rulebase is a blind taste test issue. In real life, people aren't blindfolded, and the first thing they're seeing is a profoundly different user interface, for no readily apparent reason except that you don't like the old one.

Change the colors. Put in the tabbable buttons. Waste over half the screen space with boxes and drop shadows. Suppress your gag reflex. Prove it can be done. Feel free to keep the old green menuconfig in the separate tarball with the autoconfigurator and call it "mconfig" or something, just try to make it so that when a developer runs "make menuconfig" they can't TELL whether they're running cml1 or cml2 (unless a difference they bump into is a REALLY obvious and justifiable no-brainer bugfix, like the stupid ). And the same for oldconfig and xconfig. -THAT- is far more likely to get merged.

Yes this IS a showstopper issue. Really. Honest. Stop giving objectors such an amazingly obvious target...

Rob

#120 lore raw

Of Bundling, Dao and Cowardice

In reply to Eric S. Raymond
Quoted 8 lines · Eric S. Raymond

Would you ask someone designing a new VM to make it crash and hang exactly the same way the old one did?

Do you demand that a rewrite of a disk driver have the same data-corruption bugs as the original before it can go into the tree, and tell the developer to add fixes later?

Pragmatically, the point of rewriting a system is to *fix bugs*.

That, folks, is a fine example of very common problem.
It's very tempting to try and force your ideas of How The Things Should Be Done(tm) into the tree by bundling them with genuine bug fixes and (more or less gracefully) refusing to split the patch. Anybody who had done serious work on the tree had felt that temptation at one point or another. No arguments - it's very attractive. Indeed, you don't have to defend every point of your design and implementation - you can always point to Bad Bugs(tm) that are fixed by the entire thing and obfuscate the objections away. The trouble being, _SUCH_ _BUNDLING_ _IS_ _NOT_ _A_ _VALID_ _STRATEGY_.

It had been tried. Many times. Sometimes it even got the
thing into the tree. _All_ such cases had lead to trouble.

There is another way. It takes more efforts, it requires
readiness to defend every damn part of your design and it means that you will have to deal with the sad fact that parts of that design need to be redone. It will take more time. It will look less sexy. It may very well mean that somebody else will get alternative design into the tree on the top of your efforts. There are only two positive things about it - it is honest and it leads to better tree.

How does it work? Simple - look at the path from original
tree to tree with your modifications. And no, "one big jump" doesn't count. Think of the way your ideas can be split into parts. Think of the way obstructions to the changes can be split into parts. Find small and simple changes that can go in right now, would improve the tree and would make the rest of path easier. And merge them. If you find that it can't be done - think what needs to be changed in the tree (or in your modifications) to make the split possible. If you see that something is ugly and stands in the way - help the thing to achieve the form it wants. That may change all your ideas about the right design for your modifications. That's OK - it's a Good Thing(tm). And merge the small changes. Step by step.

_That_ works. Less glory; more time; more work. And better
tree afterwards. The code wants to get cleaner. In many directions. The trick is to pick the right ones and let the thing grow into natural form. Do that many times in the right order and you will get it where you want it to be. Quite possibly - in a better place than you thought originally.

Before you start saying that it's impossible - THAT HAD BEEN DONE.
Between 2.3.40 and 2.5.2 (about two years) we got pretty much complete redesign of VFS, along with the stuff that was plain impossible with the old design. Get the old tree. Get the current tree. Compare. And realize that more than half of the way was covered during 2.4. Yes, the total size of patches had been about 2 times larger than the size of combined patch. Definitely took a lot of extra efforts. You know what? It was worth the trouble. Result is better than what I've expected. And a lot of things in the first iteration of design had turned out to be useless - stuff that was there to work around the problems that got _fixed_ by cleanups at some point during these two years. And I fully expect to see the same thing happening with the stuff I'm planning and doing now.

While we are at it, look what Rik is doing right now. He has
a huge patch pretty much rewriting (what a coincidence) VM. He had tried to shove its previous incarnations into the tree whole-sale. Saying that it didn't work is putting it _very_ mildly - resulting fight is in l-k archives and boy, was it nasty... Look what's going on now - same splitup and gradual merge strategy. Sure, extra work. Guess what? The thing had already become better from that work.

As for your motivations... If you are doing that for bragging
rights ("I've thought up a new design and now it's in the tree, exactly as I envisioned it") - find something else to brag about. Your strategy is basically a sign of cowardice - you are fighting tooth and nail to avoid gradual changes and the only way I can interpret it is that you are afraid to let parts of your design stand on their own. Not exactly a bragging material, that...

#65 lore raw

Re: Possible breakthrough in the CML2 logjam?

In reply to Dave Jones
Quoted 11 lines · Eric S. Raymond

Jeff Garzik <[email protected]>:

Quoted 2 lines

Well, let's simmer things down a bit and see what other people have to say. Maybe I'm completely off base.

Jeff, I'm not asking "other people". I'm asking *you*.

You're one of the people Linus says he trusts. Linus has said, explicitly, to myself and Dave Jones, on this very issue, "get me out of the loop". My take is that if you switch from opposing CML2 to supporting it, the political wars will probably be about over.

Shit... don't inflate my stock more than it's worth.

I'm not giving a pre-blessing to any, but I'm willing to give something an honest and fair review.

Jeff

Signature · 4 lines
Jeff Garzik      | "Why is it that attractive girls like you
Building 1024    |  always seem to have a boyfriend?"
MandrakeSoft     | "Because I'm a nympho that owns a brewery?"
                 |             - BBC TV show "Coupling"
#121 lore raw

That Linux Guy <[email protected]>,Re: Of Bundling, Dao and Cowardice

In reply to Alexander Viro
Quoted 3 lines · Alexander Viro

While we are at it, look what Rik is doing right now. He has
a huge patch pretty much rewriting (what a coincidence) VM. He had tried to shove its previous incarnations into the tree whole-sale.

Actually, Linus took it unexpectedly while I was away to Europe for conferences for 2 weeks. ;)

I don't quite agree on the "more trouble" part either, in the end it is MUCH more trouble when a big blob of code gets integrated than when a series of small changes get applied one by one.

regards,

Rik

Signature · 4 lines
"Linux holds advantages over the single-vendor commercial OS"
    -- Microsoft's "Competing with Linux" document

http://www.surriel.com/		http://distro.conectiva.com/
#125 lore raw
In reply to Eric S. Raymond
Quoted 6 lines · Eric S. Raymond

Nicolas Pitre <[email protected]>:

Quoted 2 lines

Make the whole thing ___***IDENTICAL***___ to CML1. Do a formal translation of CML1 into CML2.

Would you ask someone designing a new VM to make it crash and hang exactly the same way the old one did?

Eric, that's a slippery-slope argument. You are bright enough
to know that is not what they meant.
I like to stay out of flamewars, so I'm going to be nice,
simple, and to the point. I hack kernels. I build a lot of them. I use only make config and make oldconfig. Here is what I want from a config system (be it CML1 or CML2):
make config goes through and asks me all the questions it needs
to. The order can be different. The slight format of the prompts can be different. The backend can be wildly different. I don't CARE. What I do care about is that it only asks me relevant questions. No "How old is your computer?" Never a "Do you like the color blue?" None of that. if it wants to know my CPU, it asks "What is your cpu? [386 486 Pentium/K6 ...]" If it wants to know if I want sound support it asks "Configure sound support?" That's it. I use make config because it asks me every question, and I'm sure I haven't missed any. When it is done, I even know the macros it has set in config.h so I can peruse them in the source.
make oldconfig has to work like make old config. Again, trivial
differences in the output layout don't matter. Totally different logic dont matter. All that matters is that configuration questions I have answered already don't get asked, and questions that are new to this kernel do.
The configuration metadata should be as close to the actual item
as possible. I think everyone has come to this agreement. You know, drivers/scsi/aic_7xxx/aic_7xxx.cml or the like. Fine. I'm not even sure that I care all that much about this. So if everyone can come to a level of agreement on this, I'm sure I'll be fine.
The last thing I require is that it doesn't affect my build
time. The kernel takes a while to build right now, even on my dual-GHz machines with >GB RAM. I've not played with CML2 much, but the rumours of it taking twice as long for a build are scary. CML2 should have completed all of its work when make *config exited. All of the logic after the completion of make *config and make dep should be in the Makefiles, just like it does in CML1. Now, if I've got that wrong, let me know. I sure hope you aren't executing CML2 programs during make bzImage.
Until CML2 can come to do this, I'm going to stay perfectly
happy with CML1. mconfig just makes it easier.

Joel

Signature · 6 lines
"The first requisite of a good citizen in this republic of ours
 is that he shall be able and willing to pull his weight."
	- Theodore Roosevelt

			http://www.jlbec.org/
			[email protected]
#115 lore raw
In reply to Nicolas Pitre
Quoted 2 lines · Nicolas Pitre

Don't tell us that's not doable. Do it and show us that you can do a perfect translation of CML1 into CML2 with all CML1 structural flaws.

"Hey, the new VM in 2.4.10 should have replicated the swap overload failure case in 2.4.9! The first implementation should definitely melt down exactly the same way! We need to artificially introduce all the flaws in the old one, just to prove it can be done! Otherwise the new code is not interesting."

"To get people to try Linux on the desktop, first we need to make it blue-screen just like windows."

"It's unfair to compare laptops to desktops unless you first remove the battery from the laptop."

What the...?

Wouldn't it be nice if there was an implementation of CML2 that did everything CML1 did -EXCEPT- for the structural flaws? Rather than a blind mindless drooling bug-for-bug clone that defeats the whole purpose of reimplementing the thing?

Your requirement seems to be based on the blind assumption that CML1 had nothing whatsoever wrong with it, and CML2 didn't need to be done in the first place. If that's your argument, then say it directly. (That might be a defendable position. The one you just stated isn't.)

As for breaking CML2 so it's capable of producing a configuration that the rulebase says won't compile, the way CML1 can... You do understand the difference between a procedural and a declarative language, right?

Rob

#23 lore raw
In reply to Eric S. Raymond

Thus spake Nicolas Pitre ([email protected]):

Quoted 2 lines

As a bonus to further stimulate acceptance of CML2, make it work with the tools that most people already have i.e. Python 1.5.

I don't have python. Any version of python.

And I refuse to use software that forces me to install an ugly monstrosity like python. Integrating python in the build process is like integrating IE into Windows. It may please the stupid masses who use Red Hat anyway and don't care what happens when they click their mouse button, but it's conceptually bad for the OS and the kernel so far went to great lengths to not depend on external bloat. It is a good thing to minimize dependencies! If you can't do it in C and /bin/sh, then please step aside and let someone handle the job who is up to it.

So far this looks like you are on one hell of an ego trip.

Felix

#116 lore raw
In reply to Rob Landley
Quoted 10 lines · Rob Landley
Quoted 2 lines · Nicolas Pitre

Don't tell us that's not doable. Do it and show us that you can do a perfect translation of CML1 into CML2 with all CML1 structural flaws.

"Hey, the new VM in 2.4.10 should have replicated the swap overload failure case in 2.4.9! The first implementation should definitely melt down exactly the same way! We need to artificially introduce all the flaws in the old one, just to prove it can be done! Otherwise the new code is not interesting."

Rob:

Please read my other mails, understand the nuances, then come back, OK?

Nicolas

#61 lore raw

Re: Possible breakthrough in the CML2 logjam?

In reply to Robert Love

Thus spake Eric S. Raymond ([email protected]):

Quoted 3 lines

Jeff, I'm not asking "other people". I'm asking *you*.

You're one of the people Linus says he trusts.

Eric, why don't you try to make a system that is not openly disgusting and revolting? The kernel affects everyone. Bribing congress is a tactic worthy of Microsoft, not of you. Please concentrate on making good software and not on fast talking to Jeff until he takes a sabbatical to get some rest at all.

The more of your mails about CML I see, the more I am disgusted with it.

#85 lore raw
In reply to Alexander Viro
Quoted 3 lines · William Scott Lockwood III

again, if you all would spend 1/10th of your time helping as opposed to just slaming it, I'd be impressed. CML2 makes a WORLD of difference for me personally.

Could you elaborate on that please?

Quoted 2 lines

But hey, what do I know? I'm just a network administrator, I'm not a kernel wizard, so feel free to just totally disregard my opinions...

Signature · 1 line
Daniel
#119 lore raw
In reply to Nicolas Pitre

[ everything deleted ]

Wow, I'm amazed at the amount of time spent on this thread, when people could have actually been doing real work, or maybe just enjoying family, friends, life ...

I hope it comes to some useful resolutions,

--Matt

#71 lore raw
In reply to Eric S. Raymond
Quoted 10 lines · Michal Jaegermann

I also cannot help not to notice that the previous bit flare-up about CML2 on lkml was quelled to a great extent when somebody annouced that he is rewriting required tools in C. I had an impression that most people then shrugged "Ok, so Eric will prototope in whatever he feels comfortable with, we will have something acceptable later and we will see how this works". Now it turns out the the project got abandoned so a requirement for a huge blob of a language (as opposed to Python 1.5 which is quite smaller), which most developers do not have or need for anything else, is still there. Hm..., smells very backdoor even if it was not intended that way.

I recall there has been a complete patch set to make CML2 work with Python 1.5.2, if the two (2) in "Python 2" is your concern. The archives should tell us more.

(I'm not jumping into the other part of your mail on Python vs. Red Hat configurator bugs because I don't know that particular tool and didn't review its internal design.)

Signature · 4 lines
Matthias Andree

"They that can give up essential liberty to obtain a little temporary
safety deserve neither liberty nor safety."         Benjamin Franklin
#126 lore raw
In reply to Eric S. Raymond
Quoted 2 lines · Nicolas Pitre

Make the whole thing ___***IDENTICAL***___ to CML1. Do a formal translation of CML1 into CML2.

This requirement is contrary to CML2's objective. CML2 is not about re-implementing CML1 tools and bugs in another language, there are at least two current projects for that, one of them is mconfig.

Signature · 4 lines
Matthias Andree

"They that can give up essential liberty to obtain a little temporary
safety deserve neither liberty nor safety."         Benjamin Franklin
#33 lore raw
In reply to Larry McVoy
Quoted 5 lines · Larry McVoy
Quoted 4 lines · Matthias Andree

Are you telling that kernel programmers don't rewrite code from scratch? Is that a correct interpretation of "improve the existing system"? Note that "it can't be done" can also imply "cannot reasonable be done".

Quoted 8 lines
Quoted 2 lines

If that's not what you mean, stop reading this mail, drop a line to clarify this and forget this piece of mail.

That's not what I mean. But it's worthwhile to note that almost all "rewrite from scratch" projects really translate into "I'm unwilling to learn what the last guy did, and I'm smarter, so I'm going to do what I want to do". And that is not what kernel programmers do. They would love to be able to do that but it's very rare that doing so makes sense.

Sometimes the "old" code is set up in such a way that the "right" way to do it is not possible. That doesn't make the last guy "stupid": it can be that when he did it, that was the right way to do things, but that we've learned since then.

If the "cleanup" would result in changing more than say 60% of the code, then a rewrite is in order. The old code is usually still in place. That will allow you to "steal" code that is not the core of the
rewrite: just copy it.

I've heard that someone is trying to revolutionize "source code control systems". He's rewriting almost everything, and not enhancing any previous systems. What was his name again?

Roger.

Signature · 4 lines
** [email protected] ** http://www.BitWizard.nl/ ** +31-15-2137555 **
*-- BitWizard writes Linux device drivers for any device you may have! --*
* There are old pilots, and there are bold pilots. 
* There are also old, bald pilots. 
#60 lore raw

Re: Possible breakthrough in the CML2 logjam?

In reply to Larry McVoy
Quoted 18 lines · Larry McVoy
Quoted 8 lines · Eric S. Raymond

Larry McVoy <[email protected]>:

Quoted 7 lines
Quoted 3 lines

I need you to tell Linus that your concerns have been met and sponsor CML2 to go in, so I can stop perpetually re-fighting old battles.

That's a fine thing for anyone and everyone to say *after* they have used the system and like it.

Quoted 8 lines
Quoted 3 lines

If you are asking for a blessing in advance, which is how I read that, I would think there is zero chance of that happening, it's not how work is done on the kernel.

We're talking about design objections here. Specific objections to actual CML2 bugs, including rulebase and UI bugs, are a different level. What I am asking is if Jeff will bless the *architecture* provided the global- dependency issue is met.

Quoted 5 lines

See your quote above which contains "and sponsor CML2 to go in". Code is what goes in. Having the right architecture is great, we all agree, but what goes in is code. So your question above is basically "if I do this will you pressure Linus to accept my *code*". The answer to that should always be "no".

Sometimes I'm willing to vouch for the quality of the *code*, and I want a "go ahead and do it" from the kernel crowd.

Writing the code without consensus on the architecture can make you have to go back to architecting when people have architectural objections.

Part of the problem is that in a company the manager is in the end responsible for the salary of the guy doing the work. So he'll work along an try to make a good architecture before doing the actual coding.

For Linus it costs just one Email to say: "Hmm. Maybe. Show me the code!", and reject the code later on.

Roger.

Signature · 4 lines
** [email protected] ** http://www.BitWizard.nl/ ** +31-15-2137555 **
*-- BitWizard writes Linux device drivers for any device you may have! --*
* There are old pilots, and there are bold pilots. 
* There are also old, bald pilots. 
#34 lore raw
In reply to Larry McVoy
Quoted 9 lines · Rogier Wolff
Quoted 5 lines · Larry McVoy

That's not what I mean. But it's worthwhile to note that almost all "rewrite from scratch" projects really translate into "I'm unwilling to learn what the last guy did, and I'm smarter, so I'm going to do what I want to do". And that is not what kernel programmers do. They would love to be able to do that but it's very rare that doing so makes sense.

Sometimes the "old" code is set up in such a way that the "right" way to do it is not possible.

_Sometimes_ that is true. But there is no way you are going to be able to say that that is the typical case. History says otherwise. I'm not saying that rewrites aren't necessary, I'm saying that many, maybe most, aren't necessary.

Quoted 3 lines

I've heard that someone is trying to revolutionize "source code control systems". He's rewriting almost everything, and not enhancing any previous systems. What was his name again?

The fact that BitKeeper will read and write 30 year old SCCS files seems to have escaped your attention. Contrast that with the CML debate and I think it makes my point perfectly. If Eric's changes worked the same way, we'd have a system which worked on the old data and the new.

Signature · 2 lines
---
Larry McVoy            	 lm at bitmover.com           http://www.bitmover.com/lm 
#122 lore raw

Re: That Linux Guy <[email protected]>,Re: Of Bundling, Dao and Cowardice

In reply to Rik van Riel

In article <[email protected]> you write:

Quoted 4 lines

I don't quite agree on the "more trouble" part either, in the end it is MUCH more trouble when a big blob of code gets integrated than when a series of small changes get applied one by one.

I would say "more than one idea" here, if something need changes in a lot of places, so be it, they often won't work except as a whole. You don't cross vast chasms in a series of small safe steps.

Hanging multiple indenpendent things in a single blob is bad practice, of course.

Signature · 5 lines
bill davidsen <[email protected]> CTO, TMR Associates, Inc
  Programming without software engineering is like sculpting with a chain
saw. The very talented can produce a work of art, the mediocre wind up with
a misshapen lump in a pile of rubble, and in neither case does the end
result have more than a passing resemblance to the original intent.
#56 lore raw
In reply to Eric S. Raymond

Hi!

Quoted 3 lines

Alan, don't talk to me about "proof of concept". Tell me about a production-quality system, proven in use by people like Embedsys, Webmachines, and the Compache project. Tell me you can duplicate what

What are you trying to say? New configuration system needs to be tested by big company before you want to hear about it?

"I do not care what kernel developers say, they are all stupid, but my stuff is used by Embedsys, Webmachines and Compache, so it must be good and you are all stupid."

No wonder Linus ignores you.
Pavel

Signature · 2 lines
(about SSSCA) "I don't say this lightly.  However, I really think that the U.S.
no longer is classifiable as a democracy, but rather as a plutocracy." --hpa
#57 lore raw
In reply to Eric S. Raymond

Pavel Machek <[email protected]>:

Quoted 2 lines

What are you trying to say? New configuration system needs to be tested by big company before you want to hear about it?

No, of course not. Two of the three projects I cited are open-source developments.

My point should have been obvious. Alan is talking as though his supposed "proof of concept" is as good an argument as a tested, production-quality system. Of course, it is not.

Signature · 1 line
		<a href="http://www.tuxedo.org/~esr/">Eric S. Raymond</a>
#89 lore raw

'revolting' overemphasizes personal beliefs && religion

In reply to Dave Jones
Quoted 8 lines · Felix von Leitner

Thus spake Eric S. Raymond ([email protected]):

Quoted 3 lines

Jeff, I'm not asking "other people". I'm asking *you*.

You're one of the people Linus says he trusts.

Eric, why don't you try to make a system that is not openly disgusting and revolting? The kernel affects everyone. Bribing congress is a tactic worthy of Microsoft, not of you. Please concentrate on making

Felix, while I respect you for your work on dietlibc, calling anything that is not written in hardcore C 'revolting' is missing the point.

Quoted 1 line

The more of your mails about CML I see, the more I am disgusted with it.

And the world yawns. We do not care for (your) religious beliefs, we care *for facts*. Religion has no place in technology.

If you have an opinion, voice it without referring to how a program makes *you* feel. Voice it by stating how it should make everybody feel, based on technical arguments.

So I could say that I find your attitude disgusting, and I do, but that's not helping anybody. So instead I'll state that your arguments are counterproductive because they are only valid in the belief system of Felix von Leitner, and make no sense to others with differing belief systems.

Regards,

bert

Signature · 4 lines
http://www.PowerDNS.com          Versatile DNS Software & Services
http://www.tk                              the dot in .tk
Netherlabs BV / Rent-a-Nerd.nl           - Nerd Available -
Linux Advanced Routing & Traffic Control: http://ds9a.nl/lartc
#123 lore raw

Re: Of Bundling, Dao and Cowardice

In reply to Alexander Viro
Quoted 3 lines · Alexander Viro

How does it work? Simple - look at the path from original
tree to tree with your modifications. And no, "one big jump" doesn't count. Think of the way your ideas can be split into parts.

At least the rule base change from imperative -> inferential seems to necessarily entail a single very large atomic change, given the arch-wide rules present in the rulebase. This is the heart of the thing, and ignoring for a moment Alan's claim that it's unnecessary, it's not at all obvious that there's an incremental approach here. Any hints?

Hanging off the translation are things like the interpreter for the new language bringing it up to parity with the old interpreter, which is a meaningless delta by itself, but without which the core change is useless. The orthogonal changes are relatively minor.

Signature · 1 line
 "Love the dolphins," she advised him. "Write by W.A.S.T.E.."