Skip to contentSkip to conversation

reiser4 vs politics: linux misses out again

DISCUSSION [email protected] 20 participants 45 messages Jun 30, 2005 – Jul 11, 2005

lore
Share & tools
lore
#0 lore raw

As far as I'm concerned, commercial trolls have successfully taken away linux's only ever chance to sweep the field.

It is now gone. OSX rocks harder than linux, the spotlight function is superb.

That plus this squabbling is buying m$ enough time to make their version.

I label according to the observed effect. I haven't read the code.

Linux coulda had the OSX-type spotlight thing working, plus twice as fast filesystem 6 months or a year before Apple ... and a couple of years before m$.

Someone shoulda simply forked it then. When Hans first said 'replace VFS with reiser4'. I doubt he could have done it by himself ... they (trolls) would simply isolate his work and write his efforts off as the typical actions of a lone looney ... as they already characterise him.

The achievement reiser4 represents cannot be overstated. Genuine linux developers would have bent over backwards to get this as the primary filesystem for linux. Noone has ever doubled filesystem performance before, as far as I know, except the BeOS development team who have been divided as spoils of war between the two major competitors Apple and M$.

And he did so in a way that (a) provides for simple expansion in arbitrary directions without hackish horror such as AVFS presents; and (b) enables provision of compatibility with any arbitrary filesystem featureset.

What is all the complaining about?

And who are these guys?

Reiser has chased around trying to respond to all your criticisms for how long now? A year? Many of the critics I recall hadn't even read the material on his web site. Are these the same people now further delaying linux's adoption of 21st century filesystem semantics?

So many of the arguments I read are circular.

eg

* reiser4 can't be included until it has had widespread testing.

* reiser4 shouldn't contain two levels of plugins, since plugins properly belong in the awful hackish AVFS layer, above the VFS layer.

In fact the main impediment to reiser4 having been widely tested, in my ill-educated opinion, is simply that the directories look like files. This means a lot of application code needs minor tweaks, or at least thorough testing. Yet, it should be trivial to fix reiser4 so that directories don't look like files, no? using plugins?

Some arguments against reiser4 show that the arguer in question is even less well educated than even myself. ie, the person has not even tried reiser4.

Anyone who has, finds themselves so blown away by the apparent doubling of speed of any disk-bound task, that they start to question how much effort must have gone into making previous filesystems so slow. Who ever thought putting a transaction log at the end of the disk furthest away from where the data needs to be written would be a good thing? Why should it not go just near to where the head happens to be already? Thankyou Hans for showing us how.

To argue that benchmarks do not truly reflect real world use, you must never have even taken the time to real world use this thing. While Herr Reiser has put years of his life, and now apparently also much of his money, into creating it for you.

Try and get a wider scope on life, people. What is better for the future? Slow, dawdling development on slow dawdling filesystems and their supporting architecture, when we have been shown a fully functioning, effective and I am led to believe, simple replacement?

Well, like I said at the start, it's too late, if you do not like that path. We are now trapped there. The masses of users who would have been attracted by linux beating Apple and M$ to the punch, are now fading into some potential parallel universe. It's not this one. Hang your heads in shame, and cry. We lost. We are back to the same commercial monopoly-dominated market that we ever had. The loser is all of us, the common good, which has been sacrificed for the good of the few, by the invisible hand of the market, and the collective unconsciousness.

Simeon

#1 lore raw
Quoted 2 lines · On Thu, 30 Jun 2005 [email protected] wrote:

Reiser has chased around trying to respond to all your criticisms for how long now? A year?

The real test is in how comments are answered. Did comments on reiserfs 4 get answered by changing the code to make it more suitable for integration into Linux, or did the comments get answered only with words ?

When you're discussing code, often the best answer to criticism is a patch - not a dissertation.

Signature · 3 lines
The Theory of Escalating Commitment: "The cost of continuing mistakes is
borne by others, while the cost of admitting mistakes is borne by yourself."
  -- Joseph Stiglitz, Nobel Laureate in Economics
#2 lore raw
In reply to [email protected]
Quoted 5 lines · On Thu, 30 Jun 2005 [email protected] wrote:

As far as I'm concerned, commercial trolls have successfully taken away linux's only ever chance to sweep the field.

It is now gone. OSX rocks harder than linux, the spotlight function is superb.

[SNIPPED....]

Sorry to feed the trolls, but I couldn't resist. Isn't OSX just a commercial re-hash of BSD that Jobs "appropriated"? The performance has been so poor with the PPC platform that Apple has been forced to sign a pack with the devil and use ix86 in their future boxes.

I'm not making this up! It would have been nice to have a competing platform remaining to keep everybody honest. Now that Apple is going Intel, we will be forced to use whatever the Indians provide.

Cheers,
Dick Johnson
Penguin : Linux version 2.6.12 on an i686 machine (5537.79 BogoMips).
Notice : All mail here is now cached for review by Dictator Bush.
98.36% of all statistics are fiction.

#5 lore raw
In reply to [email protected]

Sweeping change is nice, incremental change is much more sweeter. The problem is that the "filesystem" in and of itself is where data is kept. In most cases people who run commercial businesses aren't going to change their filesystem type unless their is a clear advantage.

Therefore, you're argument negates real world. Your argument should be more about structure and a seamless migration path.

The argument isn't about technical merit.

Signature · 3 lines
.-Christopher Warner---------------------------------------------.
| my email is recycled to be sure; bcc my personal email address |
`----------------------------------------------------------------'
Quoted 88 lines · On Thu, 2005-06-30 at 22:26 +1000, [email protected] wrote:

As far as I'm concerned, commercial trolls have successfully taken away linux's only ever chance to sweep the field.

It is now gone. OSX rocks harder than linux, the spotlight function is superb.

That plus this squabbling is buying m$ enough time to make their version.

I label according to the observed effect. I haven't read the code.

Linux coulda had the OSX-type spotlight thing working, plus twice as fast filesystem 6 months or a year before Apple ... and a couple of years before m$.

Someone shoulda simply forked it then. When Hans first said 'replace VFS with reiser4'. I doubt he could have done it by himself ... they (trolls) would simply isolate his work and write his efforts off as the typical actions of a lone looney ... as they already characterise him.

The achievement reiser4 represents cannot be overstated. Genuine linux developers would have bent over backwards to get this as the primary filesystem for linux. Noone has ever doubled filesystem performance before, as far as I know, except the BeOS development team who have been divided as spoils of war between the two major competitors Apple and M$.

And he did so in a way that (a) provides for simple expansion in arbitrary directions without hackish horror such as AVFS presents; and (b) enables provision of compatibility with any arbitrary filesystem featureset.

What is all the complaining about?

And who are these guys?

Reiser has chased around trying to respond to all your criticisms for how long now? A year? Many of the critics I recall hadn't even read the material on his web site. Are these the same people now further delaying linux's adoption of 21st century filesystem semantics?

So many of the arguments I read are circular.

eg

* reiser4 can't be included until it has had widespread testing.

* reiser4 shouldn't contain two levels of plugins, since plugins properly belong in the awful hackish AVFS layer, above the VFS layer.

In fact the main impediment to reiser4 having been widely tested, in my ill-educated opinion, is simply that the directories look like files. This means a lot of application code needs minor tweaks, or at least thorough testing. Yet, it should be trivial to fix reiser4 so that directories don't look like files, no? using plugins?

Some arguments against reiser4 show that the arguer in question is even less well educated than even myself. ie, the person has not even tried reiser4.

Anyone who has, finds themselves so blown away by the apparent doubling of speed of any disk-bound task, that they start to question how much effort must have gone into making previous filesystems so slow. Who ever thought putting a transaction log at the end of the disk furthest away from where the data needs to be written would be a good thing? Why should it not go just near to where the head happens to be already? Thankyou Hans for showing us how.

To argue that benchmarks do not truly reflect real world use, you must never have even taken the time to real world use this thing. While Herr Reiser has put years of his life, and now apparently also much of his money, into creating it for you.

Try and get a wider scope on life, people. What is better for the future? Slow, dawdling development on slow dawdling filesystems and their supporting architecture, when we have been shown a fully functioning, effective and I am led to believe, simple replacement?

Well, like I said at the start, it's too late, if you do not like that path. We are now trapped there. The masses of users who would have been attracted by linux beating Apple and M$ to the punch, are now fading into some potential parallel universe. It's not this one. Hang your heads in shame, and cry. We lost. We are back to the same commercial monopoly-dominated market that we ever had. The loser is all of us, the common good, which has been sacrificed for the good of the few, by the invisible hand of the market, and the collective unconsciousness.

Simeon

-
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/

#6 lore raw
In reply to [email protected]
Quoted 2 lines · On 06/30/05 10:26:12PM +1000, [email protected] wrote:

As far as I'm concerned, commercial trolls have successfully taken away linux's only ever chance to sweep the field.

And you are?

Quoted 5 lines

It is now gone. OSX rocks harder than linux, the spotlight function is superb.

That plus this squabbling is buying m$ enough time to make their version.

MS has been working on and delaying WinFS for years, arguing with Hans for a few months will have no affect on MS' release schedule.

Quoted 3 lines

I label according to the observed effect. I haven't read the code.

Of course not, it's not like the code actually matters, right?

Quoted 3 lines

Linux coulda had the OSX-type spotlight thing working, plus twice as fast filesystem 6 months or a year before Apple ... and a couple of years before m$.

So? Most of the complaints about Linux on the desktop are userland problems. Adding cool features to the kernel won't make a big difference, if for no other reason than it will take a long time for support to make it into things like Gnome and KDE. And that's if they choose to support them, they have to support other OSes as well and adding support for features that are Linux-specific isn't to be taken lightly, especially since these would be less than Linux-specific, they would be tied to a single filesystem on Linux.

Quoted 5 lines

Someone shoulda simply forked it then. When Hans first said 'replace VFS with reiser4'. I doubt he could have done it by himself ... they (trolls) would simply isolate his work and write his efforts off as the typical actions of a lone looney ... as they already characterise him.

He can still do that, nothing is stopping him from forking Reisux and trying to woo developers.

Quoted 10 lines

The achievement reiser4 represents cannot be overstated. Genuine linux developers would have bent over backwards to get this as the primary filesystem for linux. Noone has ever doubled filesystem performance before, as far as I know, except the BeOS development team who have been divided as spoils of war between the two major competitors Apple and M$.

And he did so in a way that (a) provides for simple expansion in arbitrary directions without hackish horror such as AVFS presents; and (b) enables provision of compatibility with any arbitrary filesystem featureset.

Quoted 9 lines

What is all the complaining about?

And who are these guys?

Reiser has chased around trying to respond to all your criticisms for how long now? A year? Many of the critics I recall hadn't even read the material on his web site. Are these the same people now further delaying linux's adoption of 21st century filesystem semantics?

Probably because most of the material on his website reads like a commercial. It's hard to read through documentation when it feels like someone's constantly trying to sell you something.

Quoted 30 lines

So many of the arguments I read are circular.

eg

* reiser4 can't be included until it has had widespread testing.

* reiser4 shouldn't contain two levels of plugins, since plugins properly belong in the awful hackish AVFS layer, above the VFS layer.

In fact the main impediment to reiser4 having been widely tested, in my ill-educated opinion, is simply that the directories look like files. This means a lot of application code needs minor tweaks, or at least thorough testing. Yet, it should be trivial to fix reiser4 so that directories don't look like files, no? using plugins?

Some arguments against reiser4 show that the arguer in question is even less well educated than even myself. ie, the person has not even tried reiser4.

Anyone who has, finds themselves so blown away by the apparent doubling of speed of any disk-bound task, that they start to question how much effort must have gone into making previous filesystems so slow. Who ever thought putting a transaction log at the end of the disk furthest away from where the data needs to be written would be a good thing? Why should it not go just near to where the head happens to be already? Thankyou Hans for showing us how.

To argue that benchmarks do not truly reflect real world use, you must never have even taken the time to real world use this thing. While Herr Reiser has put years of his life, and now apparently also much of his money, into creating it for you.

Quoted 5 lines

Try and get a wider scope on life, people. What is better for the future? Slow, dawdling development on slow dawdling filesystems and their supporting architecture, when we have been shown a fully functioning, effective and I am led to believe, simple replacement?

And what is better for Linux? It's all about perspective and the people on this mailing list have to maintain the kernel from a developer's standpoint and if they start accepting every new feature regardless of complexity, maintainability, etc the kernel will become a nightmare.

And what happens in 2 years when Hans posts about reiser5 fixing all of the bad things about reiser4 and that reiser5 should be merged ASAP so that everyone can upgrade again?

Quoted 9 lines

Well, like I said at the start, it's too late, if you do not like that path. We are now trapped there. The masses of users who would have been attracted by linux beating Apple and M$ to the punch, are now fading into some potential parallel universe. It's not this one. Hang your heads in shame, and cry. We lost. We are back to the same commercial monopoly-dominated market that we ever had. The loser is all of us, the common good, which has been sacrificed for the good of the few, by the invisible hand of the market, and the collective unconsciousness.

And you're asking the kernel devs to get a wider scope on life? It sounds like you're not even living in the same reality that I am.

Quoted 2 lines

Simeon

Jim.

#7 lore raw
Quoted 2 lines · Jim Crilly
Quoted 1 line

I label according to the observed effect. I haven't read the code.

Of course not, it's not like the code actually matters, right?

As for Reiser4, they're fixing the code now to look more Linuxy and all's well.

The discussion is "Should the VFS be extended to support files-as-dirs or data objects by using what we already have in Reiser4 in -mm, although disabled."

Quoted 4 lines

So? Most of the complaints about Linux on the desktop are userland problems. Adding cool features to the kernel won't make a big difference, if for no other reason than it will take a long time for support to make it into things like Gnome and KDE. And that's if they choose to support them,

And people who'd like to use something lighter than Gnome or KDE and still use these nice features?

Quoted 4 lines

they have to support other OSes as well and adding support for features that are Linux-specific isn't to be taken lightly, especially since these would be less than Linux-specific, they would be tied to a single filesystem on Linux.

They would not be tied to a single filesystem, naturally, I think we can all agree that case is closed, as it'll just spawn another waste-of-time flamewar.

Quoted 2 lines
Quoted 2 lines

Someone shoulda simply forked it then. When Hans first said 'replace VFS with reiser4'. I doubt he could have done it by himself ... they (trolls) would

[...]

Quoted 2 lines

He can still do that, nothing is stopping him from forking Reisux and trying to woo developers.

It'd be much better to talk this thing through..
There have been pretty good arguments for the extended VFS, that it would be doable. It may just be less of a unix after that, or less of Linux as we know it now.

The circular reasoning "We don't want Reiser4's files-as-dirs in before they're tested. We also have them disabled by default. They should not be implemented on this layer here, but we won't let you touch our VFS." is bad.

Surely if the things started to go into the VFS in a separate, official tree, it'd no longer be just Namesys doing the work.

Quoted 4 lines

And what is better for Linux? It's all about perspective and the people on this mailing list have to maintain the kernel from a developer's standpoint and if they start accepting every new feature regardless of complexity, maintainability, etc the kernel will become a nightmare.

The filesystem is tested well enough to go in. For real. It may not be production stable with immediacy but it is tested.

The extended semantics are a separate matter.

Quoted 3 lines

And what happens in 2 years when Hans posts about reiser5 fixing all of the bad things about reiser4 and that reiser5 should be merged ASAP so that everyone can upgrade again?

Then someone steps up and goes "No, shut the fuck up and fix the code, we gave you your shot" or something.

Community pressure.

And it'll be a lot easier with the new VFS.

Quoted 2 lines

And you're asking the kernel devs to get a wider scope on life? It sounds like you're not even living in the same reality that I am.

Sometimes it also seems people would much rather shout at each other than see that reasons are starting to pop up why Linux could lose popularity.

I accidentally deleted the paragraph with you saying the page reads like a commercial.

I half agree, Hans has written that well, but maybe for people who would pay him money to do his work.
Therefore some of the stuff is a bit obscure. Like what is now Reiser4.1 (ie. ..metas/ whatever, I believe) is apparently referred to as Reiser6 there.

It'd be damned nice to see that page revisited a bit, maybe not much, but getting the names straight and having one of the tech guys write tech documentation that's clearly accessible.

That page does still not change the situation that the code exists to some extent, which could be merged to the VFS layer by extending it a bit and this would be easiest done in a tree that people will want to hack on.

And frankly this amount of tautology is starting to get even on my nerves :)

Signature · 1 line
mjt
#8 lore raw
Quoted 12 lines · Markus Törnqvist
Quoted 2 lines · Jim Crilly
Quoted 1 line

I label according to the observed effect. I haven't read the code.

Of course not, it's not like the code actually matters, right?

As for Reiser4, they're fixing the code now to look more Linuxy and all's well.

The discussion is "Should the VFS be extended to support files-as-dirs or data objects by using what we already have in Reiser4 in -mm, although disabled."

Quoted 8 lines
Quoted 4 lines

So? Most of the complaints about Linux on the desktop are userland problems. Adding cool features to the kernel won't make a big difference, if for no other reason than it will take a long time for support to make it into things like Gnome and KDE. And that's if they choose to support them,

And people who'd like to use something lighter than Gnome or KDE and still use these nice features?

Then work with whatever WM you want to help them support what makes sense for them. For instance, I use Enlightenment but most of the crap proposed in Reiser4 doesn't make sense for E because there's no filemanager or anything. The image thumbnails in metadata might be nice, but I wouldn't throw a fit if the current runtime generated thumbnails were still used.

Quoted 10 lines
Quoted 5 lines

they have to support other OSes as well and adding support for features that are Linux-specific isn't to be taken lightly, especially since these would be less than Linux-specific, they would be tied to a single filesystem on Linux.

They would not be tied to a single filesystem, naturally, I think we can all agree that case is closed, as it'll just spawn another waste-of-time flamewar.

Nothing has happened so far, once the VFS extensions are laid out, documented and implemented we can begin to start saying that some of the features are fs agnostic.

Quoted 11 lines
Quoted 2 lines
Quoted 2 lines

Someone shoulda simply forked it then. When Hans first said 'replace VFS with reiser4'. I doubt he could have done it by himself ... they (trolls) would

[...]

Quoted 2 lines

He can still do that, nothing is stopping him from forking Reisux and trying to woo developers.

It'd be much better to talk this thing through..
There have been pretty good arguments for the extended VFS, that it would be doable. It may just be less of a unix after that, or less of Linux as we know it now.

I'm not advocating a fork, I just think it's stupid that so many people have been saying "Stop arguing, just accept reiser4 as-is because it's fast and cool!!!!"

Quoted 8 lines

The circular reasoning "We don't want Reiser4's files-as-dirs in before they're tested. We also have them disabled by default. They should not be implemented on this layer here, but we won't let you touch our VFS." is bad.

Surely if the things started to go into the VFS in a separate, official tree, it'd no longer be just Namesys doing the work.

Quoted 8 lines
Quoted 4 lines

And what is better for Linux? It's all about perspective and the people on this mailing list have to maintain the kernel from a developer's standpoint and if they start accepting every new feature regardless of complexity, maintainability, etc the kernel will become a nightmare.

The filesystem is tested well enough to go in. For real. It may not be production stable with immediacy but it is tested.

People have been saying the same thing about reiser3 for years and yet every time I break down and try it again for whatever reason I find a nice new corner case that causes me headaches and usually ends up in me going back to ext3 or XFS.

Quoted 11 lines

The extended semantics are a separate matter.

Quoted 3 lines

And what happens in 2 years when Hans posts about reiser5 fixing all of the bad things about reiser4 and that reiser5 should be merged ASAP so that everyone can upgrade again?

Then someone steps up and goes "No, shut the fuck up and fix the code, we gave you your shot" or something.

Community pressure.

Right, because Hans is so damned receptive.

Quoted 3 lines

And it'll be a lot easier with the new VFS.

I'm not buying it, but only time will tell.

Quoted 7 lines
Quoted 2 lines

And you're asking the kernel devs to get a wider scope on life? It sounds like you're not even living in the same reality that I am.

Sometimes it also seems people would much rather shout at each other than see that reasons are starting to pop up why Linux could lose popularity.

I don't disagree, there are many aspects of each section of a Linux distro where decisions can affect how Linux is perceived by current and new users. But IMO not letting reiser4 in, in it's current state, isn't going to send a bunch of Linux users running to get Macs.

Quoted 9 lines

I accidentally deleted the paragraph with you saying the page reads like a commercial.

I half agree, Hans has written that well, but maybe for people who would pay him money to do his work.
Therefore some of the stuff is a bit obscure. Like what is now Reiser4.1 (ie. ..metas/ whatever, I believe) is apparently referred to as Reiser6 there.

I understand why he wants to have his marketing pages and I don't care if they exist, but you would think he would also want to have some SDK-type pages where it explains the new features, the "plug-in" API and such.

Quoted 9 lines

It'd be damned nice to see that page revisited a bit, maybe not much, but getting the names straight and having one of the tech guys write tech documentation that's clearly accessible.

That page does still not change the situation that the code exists to some extent, which could be merged to the VFS layer by extending it a bit and this would be easiest done in a tree that people will want to hack on.

But if big VFS changes are going to happen, I doubt anyone would want to make them in -linus or -mm because there's probably going to be a lot of initial breakage. Maybe this would be a good reason to fork 2.7?

Quoted 3 lines

And frankly this amount of tautology is starting to get even on my nerves :)

I agree, it was interesting for the first few days but now most of the threads have gone so far OT that it's just stupid.

Quoted 3 lines
Signature · 1 line
mjt

Jim.

#3 lore raw
In reply to Richard B. Johnson
Quoted 5 lines · Richard B. Johnson
Quoted 3 lines · On Thu, 30 Jun 2005 [email protected] wrote:

As far as I'm concerned, commercial trolls have successfully taken away linux's only ever chance to sweep the field.

Quoted 8 lines
Quoted 2 lines

It is now gone. OSX rocks harder than linux, the spotlight function is superb.

[SNIPPED....]

Sorry to feed the trolls, but I couldn't resist. Isn't OSX just a commercial re-hash of BSD that Jobs "appropriated"? The performance has been so poor with the PPC platform that Apple has been forced to sign a pack with the devil and use ix86 in their future boxes.

s/pack/pact/

It's not at all clear that Apple had a choice, please read comp.sys.intel for a long discussion I won't repeat.

Quoted 4 lines

I'm not making this up! It would have been nice to have a competing platform remaining to keep everybody honest. Now that Apple is going Intel, we will be forced to use whatever the Indians provide.

Or download OSX and run it elsewhere, or...

Signature · 3 lines
   -bill davidsen ([email protected])
"The secret to procrastination is to put things off until the
 last possible moment - but no longer"  -me
#44 lore raw
In reply to [email protected]
Quoted 4 lines · [email protected] writes:

As far as I'm concerned, commercial trolls have successfully taken away linux's only ever chance to sweep the field.

It is now gone. OSX rocks harder than linux, the spotlight function is superb.

Speaking of trolls...

-Miles

Signature · 1 line
Suburbia: where they tear out the trees and then name streets after them.
#4 lore raw
In reply to Richard B. Johnson
Quoted 6 lines · Richard B. Johnson

Sorry to feed the trolls, but I couldn't resist. Isn't OSX just a commercial re-hash of BSD that Jobs "appropriated"? The performance has been so poor with the PPC platform that Apple has been forced to sign a pack with the devil and use ix86 in their future boxes.

Actually, OSX is more like that old and superb NEXTSTEP.

#9 lore raw
In reply to Jim Crilly
Quoted 10 lines · Jim Crilly
Quoted 5 lines · Markus Törnqvist

It'd be much better to talk this thing through..
There have been pretty good arguments for the extended VFS, that it would be doable. It may just be less of a unix after that, or less of Linux as we know it now.

I'm not advocating a fork, I just think it's stupid that so many people have been saying "Stop arguing, just accept reiser4 as-is because it's fast and cool!!!!"

Assuming "fast and cool" here equates to some level of improvement to the existing kernel, and/or new features/capabilities not currently available, then are you saying "fast and cool" isn't good enough anymore, you now have to be politically correct and socially popular and a master brown-noser as well to get code into the kernel even just on an *experimental* basis?

In reality, the implied attitude behind your statement actually *guarantees* a fork of Linux at some point down the road if you keep stonewalling the inclusion of something that clearly has enormous potential, because for many people "fast and cool" IS THE DESIRED OBJECTIVE, and by saying no to that, YOU are the one setting the stage for a fork. R4, or any other promising and relatively stable technology, can't reach its full potential until it begins to see wider testing and usage, which can only happen when it gets included into the mainstream stable kernel. So when the fork finally happens because people got tired of waiting, don't blame the forkers and don't blame Hans, it will have been your own attitude that made it a fait accompli.

You obviously don't realize it, but a lot of us don't give a damn whether you like Hans or not, or whether you think his design is "clean" or "correct" or not, we just want Linux to be the "fastest and coolest" it can be. Find a way to solve the technical problems, and just give R4 its shot. Let it succeed or fail based on its own merits, and not because its author may have poor social skills, or simply rubs you the wrong way.

I hope your attitude turns out to be in the minority, and cooler heads prevail in this argument, because otherwise, Unix historians will likely look back to this moment and say this was when Linux began to stumble...

#10 lore raw
Quoted 11 lines · Ed Cogburn

Assuming "fast and cool" here equates to some level of improvement to the existing kernel, and/or new features/capabilities not currently available, then are you saying "fast and cool" isn't good enough anymore, you now have to be politically correct and socially popular and a master brown-noser as well to get code into the kernel even just on an *experimental* basis?

In reality, the implied attitude behind your statement actually *guarantees* a fork of Linux at some point down the road if you keep stonewalling the inclusion of something that clearly has enormous potential, because for many people "fast and cool" IS THE DESIRED OBJECTIVE, and by saying no to that, YOU are the one setting the stage for a fork.

In fact such a fork would be a great thing. Let people play fast and loose who want to do that, and if they don't manage to break it and the feature is still cool a while later we clean it up and merge it into Linux proper.

#12 lore raw
In reply to Ed Cogburn
Quoted 14 lines · Ed Cogburn
Quoted 10 lines · Jim Crilly
Quoted 5 lines · Markus Törnqvist

It'd be much better to talk this thing through..
There have been pretty good arguments for the extended VFS, that it would be doable. It may just be less of a unix after that, or less of Linux as we know it now.

I'm not advocating a fork, I just think it's stupid that so many people have been saying "Stop arguing, just accept reiser4 as-is because it's fast and cool!!!!"

Quoted 6 lines

Assuming "fast and cool" here equates to some level of improvement to the existing kernel, and/or new features/capabilities not currently available, then are you saying "fast and cool" isn't good enough anymore, you now have to be politically correct and socially popular and a master brown-noser as well to get code into the kernel even just on an *experimental* basis?

Fast and cool by themselves shouldn't be good enough, infact fast could definately be optional because if the implementation is good and the code is clean it will get optimized down the road. And IMO reiser4 is an exception to the "but just mark it experimental" argument because it's so large and implements so many things that should be in other places. If it was something small like a device driver that was in 1-5 files and didn't touch anything else, I'm sure no one would argue against it's inclusion.

Quoted 12 lines

In reality, the implied attitude behind your statement actually *guarantees* a fork of Linux at some point down the road if you keep stonewalling the inclusion of something that clearly has enormous potential, because for many people "fast and cool" IS THE DESIRED OBJECTIVE, and by saying no to that, YOU are the one setting the stage for a fork. R4, or any other promising and relatively stable technology, can't reach its full potential until it begins to see wider testing and usage, which can only happen when it gets included into the mainstream stable kernel. So when the fork finally happens because people got tired of waiting, don't blame the forkers and don't blame Hans, it will have been your own attitude that made it a fait accompli.

Quoted 7 lines

You obviously don't realize it, but a lot of us don't give a damn whether you like Hans or not, or whether you think his design is "clean" or "correct" or not, we just want Linux to be the "fastest and coolest" it can be. Find a way to solve the technical problems, and just give R4 its shot. Let it succeed or fail based on its own merits, and not because its author may have poor social skills, or simply rubs you the wrong way.

Obviously you don't care about the code or it's maintainer's attitude, you don't have to deal with either of them. And if you had read all of the reiser4 threads you would have seen that people are trying to work out the technical details, it's not a fast process when 2 things this large and complicated are involved.

Quoted 4 lines

I hope your attitude turns out to be in the minority, and cooler heads prevail in this argument, because otherwise, Unix historians will likely look back to this moment and say this was when Linux began to stumble...

I'm glad the kernel maintaners are being convservative about what gets merged, if they allowed everything in that was sent to lkml the historians would be looking back and saying this was when Linux started to become unmaintainable.

Jim.

#11 lore raw
In reply to Christoph Hellwig
Quoted 15 lines · Christoph Hellwig
Quoted 7 lines · Ed Cogburn

Assuming "fast and cool" here equates to some level of improvement to the existing kernel, and/or new features/capabilities not currently available, then are you saying "fast and cool" isn't good enough anymore, you now have to be politically correct and socially popular and a master brown-noser as well to get code into the kernel even just on an *experimental* basis?

Quoted 6 lines

In reality, the implied attitude behind your statement actually *guarantees* a fork of Linux at some point down the road if you keep stonewalling the inclusion of something that clearly has enormous potential, because for many people "fast and cool" IS THE DESIRED OBJECTIVE, and by saying no to that, YOU are the one setting the stage for a fork.

Quoted 2 lines

In fact such a fork would be a great thing. Let people play fast and loose who want to do that,

Here's the irony of your statement: -mm is too unstable for many people to even try out R4 with, breaking or freezing or having other problems long before they even get to R4 itself.

Please remember that the comment was "fast and cool", not "fast and unstable".

#13 lore raw
In reply to Jim Crilly
Quoted 10 lines · Jim Crilly
Quoted 8 lines · Ed Cogburn

Jim Crilly wrote:

Assuming "fast and cool" here equates to some level of improvement to the existing kernel, and/or new features/capabilities not currently available, then are you saying "fast and cool" isn't good enough anymore, you now have to be politically correct and socially popular and a master brown-noser as well to get code into the kernel even just on an *experimental* basis?

Quoted 2 lines

Fast and cool by themselves shouldn't be good enough, infact fast could definately be optional

And please point out where anyone has said ANYWHERE that R4 is going to be FORCED on people! Stop accusing Hans of refusing to answer relevent points when you're doing the same damn thing.

Quoted 1 line

Obviously you don't care about the code

Bullshit. If the code doesn't work or is unstable or unmaintained then everyone is concerned. BUT THOSE ARE NOT THE ISSUES HERE. Maybe Hans will be a little more sociable towards you when you guys start being a little more sociable towards him, or maybe he's just a jerk and will never change, I don't know, and frankly I don't care, but R4 deserves its shot anyway based just on its potential alone. Its reached the point where it can only get better with more *exposure*, and as I said in my response to Christoph, -mm is not the place to get exposure from a wider audience, due to its instability, even Andrew has said his -mm tree isn't meant to be a "stable" kernel series.

In the meantime, if a fork does happen, I'm going with the one that gives good ideas their chance at the mainstream no matter who the author is. If you guys really want to kill the Linux success story, just keep right on putting your ego and personal feelings in front of what's good for the users and good for the success of Linux outside your elite club. And before you tell yourself that the users don't matter, you'd better ask around among all the *other* developers here why *they* are working on Linux now, and not a BSD, or another fork of Linux. Hint: many are employed by companies whose customer's are ... Linux *USERS*. Linux's initial success was not pre-ordained or guaranteed, it was accomplished because a core group of people started WORKING TOGETHER with a common vision, if you now decide to forget what got you to this point, I GUARANTEE YOU that the success of your version of Linux will not continue.

I know there are technical issues, in a few places by a few people, like the thread Andrew started, they are still being discussed, but when we reach the point where people say "fast and cool" is not good or desirable anymore, without realizing how absurd that sounds, then the dispute has clearly gone beyond the technical, and into the realm of the ego stratosphere. Believe it or not Jim, but I'm absolutely sure Hans really wants whats best for Linux in the long run just as you do. R4 deserves its chance, so find a way to make it work, and that goes to Hans too.

Thank you all for bearing through this rant...

Flame retardant attire now in place, you may fire at your convenience gentlemen. :)

#14 lore raw
Quoted 16 lines · Ed Cogburn
Quoted 10 lines · Jim Crilly
Quoted 8 lines · Ed Cogburn

Jim Crilly wrote:

Assuming "fast and cool" here equates to some level of improvement to the existing kernel, and/or new features/capabilities not currently available, then are you saying "fast and cool" isn't good enough anymore, you now have to be politically correct and socially popular and a master brown-noser as well to get code into the kernel even just on an *experimental* basis?

Quoted 2 lines

Fast and cool by themselves shouldn't be good enough, infact fast could definately be optional

Quoted 6 lines

And please point out where anyone has said ANYWHERE that R4 is going to be FORCED on people! Stop accusing Hans of refusing to answer relevent points when you're doing the same damn thing.

From a user's standpoint, no of course not. But once it's included in the

mainline it's forced upon the kernel maintainers and if parts of it are used to extend the VFS it's forced upon even more of the kernel maintainers.

And when have I ever accused Hans of anything?

Quoted 4 lines
Quoted 2 lines

Obviously you don't care about the code

Quoted 11 lines

Bullshit. If the code doesn't work or is unstable or unmaintained then everyone is concerned. BUT THOSE ARE NOT THE ISSUES HERE. Maybe Hans will be a little more sociable towards you when you guys start being a little more sociable towards him, or maybe he's just a jerk and will never change, I don't know, and frankly I don't care, but R4 deserves its shot anyway based just on its potential alone. Its reached the point where it can only get better with more *exposure*, and as I said in my response to Christoph, -mm is not the place to get exposure from a wider audience, due to its instability, even Andrew has said his -mm tree isn't meant to be a "stable" kernel series.

Yes, they are the issues. But I believe most of the big problems have been fixed (i.e. things like R4 containing it's own hashing library instead of using what's already in the kernel) in the original thread. And from what I remember of the R3 saga back right after the 2.4.0 release, this looks extremely familiar so I really doubt Hans will be changing any time soon.

And I agree that -mm is too unstable for R4 to get serious testing there, the few times I gave it a try I always had more than one problem that forced me back to -linus with a few custom patches.

Quoted 5 lines

In the meantime, if a fork does happen, I'm going with the one that gives good ideas their chance at the mainstream no matter who the author is. If you guys really want to kill the Linux success story, just keep right on putting your ego and personal feelings in front of what's good for the users and good for the success of Linux outside your elite club. And

Quoted 9 lines

before you tell yourself that the users don't matter, you'd better ask around among all the *other* developers here why *they* are working on Linux now, and not a BSD, or another fork of Linux. Hint: many are employed by companies whose customer's are ... Linux *USERS*. Linux's initial success was not pre-ordained or guaranteed, it was accomplished because a core group of people started WORKING TOGETHER with a common vision, if you now decide to forget what got you to this point, I GUARANTEE YOU that the success of your version of Linux will not continue.

I think you're putting too much of an idealogical spin on the history of Linux. And I seem to remember a lot of people claiming that R3 was the filesystem of the future back around it's release as well and as far as I can tell there's only a few major distributions that support R3 at all.

If R4 isn't included it won't kill anything and if anything these discussions have already gotten people thinking about new things that can be done at the VFS level, so I wouldn't be surprised if we see VFS plugins to handle things like on the fly compression whether R4 is included or not.

Quoted 9 lines

I know there are technical issues, in a few places by a few people, like the thread Andrew started, they are still being discussed, but when we reach the point where people say "fast and cool" is not good or desirable anymore, without realizing how absurd that sounds, then the dispute has clearly gone beyond the technical, and into the realm of the ego stratosphere. Believe it or not Jim, but I'm absolutely sure Hans really wants whats best for Linux in the long run just as you do. R4 deserves its chance, so find a way to make it work, and that goes to Hans too.

I still believe that "fast and cool" alone isn't enough. If someone were to offer to sell you car that looked extremely cool and was outrageously fast to the tune of like 2000hp would you buy it without thinking or would you want to make sure it had other important features like anti-lock breaks, airbags and someone who understands the engine enough to fix it when something gets broke?

Quoted 4 lines

Thank you all for bearing through this rant...

Flame retardant attire now in place, you may fire at your convenience gentlemen. :)

Jim.

#15 lore raw
Quoted 2 lines · Jim Crilly

From a user's standpoint, no of course not. But once it's included in the mainline it's forced upon the kernel maintainers

No it isn't. Its just like any other optional feature. If it stays unmaintained for too long it just gets removed from the kernel, this happens all the time. This is another one of those non-issues that isn't relevent to this discussion but is being used to avoid the real issue. Go waste someone else's time, Jim.

#16 lore raw
Quoted 1 line · Ed Cogburn

Jim Crilly wrote:

[...]

Quoted 6 lines
Quoted 2 lines

From a user's standpoint, no of course not. But once it's included in the mainline it's forced upon the kernel maintainers

No it isn't. Its just like any other optional feature. If it stays unmaintained for too long it just gets removed from the kernel, this happens all the time.

The work involved in getting it into the kernel, workin with it while it is there, and then ripping it out when it becomes clear that it won't be maintained anymore doesn't count.

Quoted 2 lines

This is another one of those non-issues that isn't
relevent to this discussion but is being used to avoid the real issue.

How to allocate resources /is/ the central issue in any software development enterprise. That this is open source makes it somewhat different, in that most resources are spent integrating stuff developed ouside, and maintaining it later.

Quoted 2 lines

Go
waste someone else's time, Jim.

Good advise.

Signature · 4 lines
Dr. Horst H. von Brand                   User #22616 counter.li.org
Departamento de Informatica                     Fono: +56 32 654431
Universidad Tecnica Federico Santa Maria              +56 32 654239
Casilla 110-V, Valparaiso, Chile                Fax:  +56 32 797513
#43 lore raw
In reply to Ed Cogburn
Quoted 11 lines · Ed Cogburn
Quoted 9 lines · Jim Crilly
Quoted 8 lines · Ed Cogburn

Jim Crilly wrote:

Assuming "fast and cool" here equates to some level of improvement to the existing kernel, and/or new features/capabilities not currently available, then are you saying "fast and cool" isn't good enough anymore, you now have to be politically correct and socially popular and a master brown-noser as well to get code into the kernel even just on an *experimental* basis?

Quoted 6 lines
Quoted 2 lines

Fast and cool by themselves shouldn't be good enough, infact fast could definately be optional

And please point out where anyone has said ANYWHERE that R4 is going to be FORCED on people! Stop accusing Hans of refusing to answer relevent points when you're doing the same damn thing.

Even if I'm not forced to /use/ it, if it gets shoved into the kernel, it is being forced on me. You are talking to /kernel developers/ here, who care for the /code/; not /users/, who only care for whatever their distribution dishes up (leaving out experimental/unstable/nonsensical stuff).

Quoted 3 lines
Quoted 1 line

Obviously you don't care about the code

Bullshit. If the code doesn't work

Check. Can't, due to basic design problems.

Quoted 1 line

or is unstable

Check.

Quoted 1 line

or unmaintained

Has happened before with code from the same people. Can't even get them to clean up for considering for inclusion. Check.

Quoted 2 lines

then
everyone is concerned.

Yep.

Quoted 3 lines

BUT THOSE ARE NOT THE ISSUES HERE. Maybe Hans will
be a little more sociable towards you when you guys start being a little more sociable towards him,

Sorry, the community has been sometimes abrasive, but on the whole helpful and polite to newcomers. In this case they have been accused of everything, including dishonesty... and this has been going on for the 4th round right now.

Quoted 1 line

or maybe he's just a jerk and will never

change,

Looks that way. And /again/ there are people who tell nicely what would have to happen to fix ReiserFS 4 to get it into shape.

Quoted 2 lines

I don't know, and frankly I don't care, but R4 deserves its shot anyway based just on its potential alone.

And who will take over when the authors make for the woodwork, yet again?

Quoted 5 lines

Its reached the point where it can only
get better with more *exposure*, and as I said in my response to Christoph, -mm is not the place to get exposure from a wider audience, due to its instability, even Andrew has said his -mm tree isn't meant to be a "stable" kernel series.

And what stops you from creating -rfs4 just for checking? If it /really/ is so great, word will go round and people /will/ check it out.

Quoted 5 lines

In the meantime, if a fork does happen, I'm going with the one that gives good ideas their chance at the mainstream no matter who the author is. If you guys really want to kill the Linux success story, just keep right on putting your ego and personal feelings in front of what's good for the users and good for the success of Linux outside your elite club.

Addressed at the ReiserFS crowd? Sounds about right...

Quoted 4 lines

And
before you tell yourself that the users don't matter, you'd better ask around among all the *other* developers here why *they* are working on Linux now, and not a BSD, or another fork of Linux.

A /very/ good question...

Quoted 7 lines

Hint: many are
employed by companies whose customer's are ... Linux *USERS*. Linux's initial success was not pre-ordained or guaranteed, it was accomplished because a core group of people started WORKING TOGETHER with a common vision, if you now decide to forget what got you to this point, I GUARANTEE YOU that the success of your version of Linux will not continue.

And again, historically "Work towards /my/ vision", "Linux is doomed if you don't do as I say", etc arguments go nowhere. Rightly so, they are usually from people who don't have a clue. If you /really/ can do it /that/ much better than Linus & Co, go ahead and show us.

Signature · 4 lines
Dr. Horst H. von Brand                   User #22616 counter.li.org
Departamento de Informatica                     Fono: +56 32 654431
Universidad Tecnica Federico Santa Maria              +56 32 654239
Casilla 110-V, Valparaiso, Chile                Fax:  +56 32 797513
#41 lore raw
In reply to Ed Cogburn
Quoted 10 lines · Ed Cogburn
Quoted 2 lines · Jim Crilly

From a user's standpoint, no of course not. But once it's included in the mainline it's forced upon the kernel maintainers

No it isn't. Its just like any other optional feature. If it stays unmaintained for too long it just gets removed from the kernel, this happens all the time. This is another one of those non-issues that isn't relevent to this discussion but is being used to avoid the real issue. Go

Yeah right, you really nailed that one, Ed. You obviously have years of kernel maintenance under your belt.

Signature · 1 line
Jens Axboe
#42 lore raw
Quoted 3 lines · Ed Cogburn

Here's the irony of your statement: -mm is too unstable for many people to even try out R4 with, breaking or freezing or having other problems long before they even get to R4 itself.

I remember that Con Kolivas had r4 in his -ck patchset for a bit.

He dropped it, but a fair amount of people is using -ck as it is stable, so maybe, if that is not too much for Con, he could merge r4 with his patchset or create an additional -ck-r4

Just a vague thought.

Waddya think? Con?

Regards,
Maciej

#17 lore raw
In reply to Horst von Brand

----- Original Message -----
From: "Horst von Brand" <[email protected]>
To: "Ed Cogburn" <[email protected]>
Cc: <[email protected]>
Sent: Sunday, July 03, 2005 22:11
Subject: Re: reiser4 vs politics: linux misses out again

Quoted 2 lines

The work involved in getting it into the kernel, workin with it while it

is

Quoted 2 lines

there, and then ripping it out when it becomes clear that it won't be maintained anymore doesn't count.

In reality is it doesn't count. Users don't care what level of pain is involved in producing the products they use.

Development efforts and results for OS's are always just taken for granted.

BTDT - if you're very lucky, a (very) few non-programming users might notice something nice and mention that they noticed a difference. The majority are still struggling to find the power switch ;->

#18 lore raw
Quoted 10 lines · [email protected] wrote:

In reality is it doesn't count. Users don't care what level of pain is involved in producing the products they use.

Development efforts and results for OS's are always just taken for granted.

BTDT - if you're very lucky, a (very) few non-programming users might notice something nice and mention that they noticed a difference. The majority are still struggling to find the power switch ;->

You no longer have to be a kernel dev to see that there is more to the resistance to R4 than objective technical issues, anyone with an understanding of English whose been reading the R4 debates-that-quickly-turn-into-flame-wars the last couple of years here can see that. For you guys to continue to suggest otherwise only makes you out to be the fools, not the "lusers" (which you obviously define as anyone who isn't a kernel dev).

So be my guest, ignore the message and attack the messenger, I didn't respond to start yet another flamewar, nor did I really expect much objectivity anyway, as that's been thrown out the window even in discussions between developers, e.g. the R4 plugins thread.

If its a fork of the kernel that you really want, so be it. When it happens, and given the increasing divergence going on between the commercial distros and the vanilla kernel, maybe it's already begun, I'll use the one that isn't afraid of giving new ideas a chance.

Flame away, I'm done.

#19 lore raw
Quoted 26 lines · Ed Cogburn
Quoted 10 lines · [email protected] wrote:

In reality is it doesn't count. Users don't care what level of pain is involved in producing the products they use.

Development efforts and results for OS's are always just taken for granted.

BTDT - if you're very lucky, a (very) few non-programming users might notice something nice and mention that they noticed a difference. The majority are still struggling to find the power switch ;->

You no longer have to be a kernel dev to see that there is more to the resistance to R4 than objective technical issues, anyone with an understanding of English whose been reading the R4 debates-that-quickly-turn-into-flame-wars the last couple of years here can see that. For you guys to continue to suggest otherwise only makes you out to be the fools, not the "lusers" (which you obviously define as anyone who isn't a kernel dev).

So be my guest, ignore the message and attack the messenger, I didn't respond to start yet another flamewar, nor did I really expect much objectivity anyway, as that's been thrown out the window even in discussions between developers, e.g. the R4 plugins thread.

Quoted 6 lines

If its a fork of the kernel that you really want, so be it. When it happens, and given the increasing divergence going on between the commercial distros and the vanilla kernel, maybe it's already begun, I'll use the one that isn't afraid of giving new ideas a chance.

Flame away, I'm done.

No Flame from me. One thing to remember is that Hans and friends _have_ supported R3 for years. This is an undisputed fact. Second third parties have be able to add much function (like journaling) to R3 so the code must be sort of readable... With R4 they have created a new FS that is _fast_ and _can_ do things no other FS can - I also expect they have written cleaner code... Why are we fighting about adding this sort of function to the kernel? Yes it may not be the absolute best way to do things. How many times has tcpip be rewritten for linux? The answer is more than once. Lets put R4 in, see how it works, generalize the ideas and if we have to rewrite and rethink part of it lets do so.

Please do add R4 to the kernel!

Thanks,

Ed Tomlinson

#20 lore raw
Quoted 9 lines · Ed Tomlinson

No Flame from me. One thing to remember is that Hans and friends _have_ supported R3 for years. This is an undisputed fact. Second third parties have be able to add much function (like journaling) to R3 so the code must be sort of readable... With R4 they have created a new FS that is _fast_ and _can_ do things no other FS can - I also expect they have written cleaner code... Why are we fighting about adding this sort of function to the kernel? Yes it may not be the absolute best way to do things. How many times has tcpip be rewritten for linux? The answer is more than once. Lets put R4 in, see how it works, generalize the ideas and if we have to rewrite and rethink part of it lets do so.

remember that Hans is on record (over a year ago) arguing that R3 should not be fixed becouse R4 was replacing it.

This type of thing is one of the reasons that you see arguments that aren't 'purely code-related' becouse the kernel folks realize that _they_ will have to maintain the code over time, Hans and company will go on and develop R5 (R10, whatever) and consider R4 obsolete and stop maintaining it.

David Lang

Signature · 2 lines
There are two ways of constructing a software design. One way is to make it so simple that there are obviously no deficiencies. And the other way is to make it so complicated that there are no obvious deficiencies.
 -- C.A.R. Hoare
#21 lore raw
Quoted 18 lines · David Lang
Quoted 15 lines · Ed Tomlinson

No Flame from me. One thing to remember is that Hans and friends _have_ supported
R3 for years. This is an undisputed fact. Second third parties have be able to add much
function (like journaling) to R3 so the code must be sort of readable... With R4 they have created a new FS that is _fast_ and _can_ do things no other FS can - I also expect they have
written cleaner code... Why are we fighting about adding this sort of function to the kernel?
Yes it may not be the absolute best way to do things. How many times has tcpip be rewritten
for linux? The answer is more than once. Lets put R4 in, see how it works, generalize the ideas and if we have to rewrite and rethink part of it lets do so.

Quoted 8 lines

remember that Hans is on record (over a year ago) arguing that R3 should not be fixed becouse R4 was replacing it.

This type of thing is one of the reasons that you see arguments that aren't 'purely code-related' becouse the kernel folks realize that _they_ will have to maintain the code over time, Hans and company will go on and develop R5 (R10, whatever) and consider R4 obsolete and stop maintaining it.

Maybe its because Hans and Co., having only a finite amount of dev time, would much prefer to spend that time on R4 rather than R3? Maybe if we were to let R4 into the kernel, it wouldn't be long after that R3 could be retired because everyone has moved to R4? Maybe if R4 had been allowed into the mainstream kernel a year ago, this ENTIRE ARGUMENT would be moot?

Since when has there been a requirement for all new code to have a signed support contract before it enters the kernel? Don't tell me this is normal, because I damn well know its not. Early on and for the most part even today, good stuff gets into the kernel without any explicit or at least meaningful promises of long term support. That stuff stays in the kernel as long as there is someone willing to maintain it, and it disappears from the kernel when there is no one left willing to keep it up-to-date. That has always been the basic rule, without it Linux would never have gone beyond the stage of being Linus's toy, because if Linus had demanded such a support commitment up front, people would have lost interest in hacking on his toy.

Devs should be free to work on whatever they want, because most of them are doing this on their own time anyway, otherwise they might just decide to hack on some other OS, or a fork of Linux instead. Demanding long term support commitment from volunteers is just bizarre, since pretty soon, you won't have many volunteers left, with the ones still remaining consistently lying to you as everyone will know that the promises they are making can't be enforced..... As for code from companies, has Linus gotten a signed support contract from IBM for JFS? If he asked, do you really think they would agree to one?

So this business of demanding a perpetual support contract from Hans for R3 before you even let R4 into the kernel is patently absurd. Such a contract won't be needed once R4 gets into widespread use and stabilizes, allowing R3 users to switch over to it, never mind that this has not been required to my knowledge of anyone else. Either you're putting the cart before the horse on this one without realizing it, or you're just another member of the 'say-no-to-change' group (or worse, the say-no-to-Hans group) using any excuse that pops into your head for why you keep saying 'no'.

#40 lore raw
In reply to David Lang
Quoted 3 lines · David Lang

remember that Hans is on record (over a year ago) arguing that R3 should not be fixed becouse R4 was replacing it.

No, I said and say that V3 should not have features added to it, because features should not be added to a stable branch. Bug fixes are good.

There are a few V3 bug fixes where the fix is so deep that it belongs in V4, all of the ones that I can think of at the moment are ones requiring disk format changes.

Note that in V4, disk format changes are no longer deep fixes because of plugins.

Quoted 6 lines

This type of thing is one of the reasons that you see arguments that aren't 'purely code-related' becouse the kernel folks realize that _they_ will have to maintain the code over time, Hans and company will go on and develop R5 (R10, whatever) and consider R4 obsolete and stop maintaining it.

No, we will stop adding features to it at some point, only add bug fixes, and let it become stable enough for mission critical use. Of course, with plugins this becomes more complicated of a policy because smaller releases with more orthogonal features are easier and more tempting, and it becomes tempting to version and release plugins rather than the FS, so I am not sure exactly how this will play out yet. I think we will have an option to select experimental plugins individually.

Quoted 3 lines

David Lang

#22 lore raw
In reply to Ed Cogburn
Quoted 28 lines · Ed Cogburn

Maybe its because Hans and Co., having only a finite amount of dev time, would much prefer to spend that time on R4 rather than R3? Maybe if we were to let R4 into the kernel, it wouldn't be long after that R3 could be retired because everyone has moved to R4? Maybe if R4 had been allowed into the mainstream kernel a year ago, this ENTIRE ARGUMENT would be moot?

Since when has there been a requirement for all new code to have a signed support contract before it enters the kernel? Don't tell me this is normal, because I damn well know its not. Early on and for the most part even today, good stuff gets into the kernel without any explicit or at least meaningful promises of long term support. That stuff stays in the kernel as long as there is someone willing to maintain it, and it disappears from the kernel when there is no one left willing to keep it up-to-date. That has always been the basic rule, without it Linux would never have gone beyond the stage of being Linus's toy, because if Linus had demanded such a support commitment up front, people would have lost interest in hacking on his toy.

Devs should be free to work on whatever they want, because most of them are doing this on their own time anyway, otherwise they might just decide to hack on some other OS, or a fork of Linux instead. Demanding long term support commitment from volunteers is just bizarre, since pretty soon, you won't have many volunteers left, with the ones still remaining consistently lying to you as everyone will know that the promises they are making can't be enforced..... As for code from companies, has Linus gotten a signed support contract from IBM for JFS? If he asked, do you really think they would agree to one?

Quoted 9 lines

So this business of demanding a perpetual support contract from Hans for R3 before you even let R4 into the kernel is patently absurd. Such a contract won't be needed once R4 gets into widespread use and stabilizes, allowing R3 users to switch over to it, never mind that this has not been required to my knowledge of anyone else. Either you're putting the cart before the horse on this one without realizing it, or you're just another member of the 'say-no-to-change' group (or worse, the say-no-to-Hans group) using any excuse that pops into your head for why you keep saying 'no'.

it's not that they are being asked to commit to a perpetual support contract, it's the fact that they AREN'T being asked to commit to a perpetual support contract that makes it more important that the code that they are asking to be merged must fit in well with the code and the philosphy of the rest of the kernel so that it can be supported well by others

if the kernel folks don't agree with the underlying design they may end up accepting it for a while, but will rip it out as soon as they get an excuse (see devfs for a perfect example)

David Lang

--
There are two ways of constructing a software design. One way is to make it so simple that there are obviously no deficiencies. And the other way is to make it so complicated that there are no obvious deficiencies.
-- C.A.R. Hoare

#23 lore raw
In reply to Ed Cogburn
Quoted 5 lines · Ed Cogburn
Quoted 4 lines · David Lang
Quoted 2 lines · Ed Tomlinson

No Flame from me. One thing to remember is that Hans and friends _have_ supported R3 for years.

They let it fall into disrepair when they started work on 4.

Quoted 1 line
Quoted 1 line
Quoted 1 line

This is an undisputed fact.

Exactly.

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

Second
third parties have be able to add much function (like journaling) to R3 so the code must be sort of readable...

Why don't you check it? Wouldn't you much prefer if the original authors (or somebody similarly initmate with the code) did mayor surgery on it? Specially if it is something you depend on?

Quoted 1 line
Quoted 1 line
Quoted 1 line

With R4 they have created a new FS that is _fast_

Remains to be seen.

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

and _can_ do things
no other FS can

Mostly useless things...

Quoted 1 line
Quoted 1 line
Quoted 1 line

- I also expect they have written cleaner code...

Better check first.

Quoted 1 line
Quoted 1 line
Quoted 1 line

Why are we fighting about adding this sort of function to the kernel?

Because the filessytem experts in the kernel development crowd (and others) have /serious/ problems with the ideas and the code?

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Yes it may not be the absolute best way to do things. How many times has tcpip be rewritten for linux? The answer is more than once.

So?

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Lets put R4 in, see how it works, generalize the ideas and if we have to rewrite and rethink part of it lets do so.

Why not: Let's keep it out, fix the problems that it has and evaluate it for inclusion once the problems have been ironed out? That has been the policy for everything else as far as I can remember (and that is from nearly the beginning...)

Quoted 11 lines
Quoted 8 lines

remember that Hans is on record (over a year ago) arguing that R3 should not be fixed becouse R4 was replacing it.

This type of thing is one of the reasons that you see arguments that aren't 'purely code-related' becouse the kernel folks realize that _they_ will have to maintain the code over time, Hans and company will go on and develop R5 (R10, whatever) and consider R4 obsolete and stop maintaining it.

Maybe its because Hans and Co., having only a finite amount of dev time, would much prefer to spend that time on R4 rather than R3?

ext2 is still being maintained alongside ext3.

Quoted 3 lines

Maybe if we
were to let R4 into the kernel, it wouldn't be long after that R3 could be retired because everyone has moved to R4?

ext3 is several years old, and there are /still/ ext2 users around...

[...]

Quoted 3 lines

Devs should be free to work on whatever they want, because most of them are doing this on their own time anyway, otherwise they might just decide to hack on some other OS, or a fork of Linux instead.

Nobody forces anybody to work on Linux, or even on the standard Linus kernel. It is the ReiserFS crowd who are demanding something from the Linux crowd, not the other way around.

Signature · 4 lines
Dr. Horst H. von Brand                   User #22616 counter.li.org
Departamento de Informatica                     Fono: +56 32 654431
Universidad Tecnica Federico Santa Maria              +56 32 654239
Casilla 110-V, Valparaiso, Chile                Fax:  +56 32 797513
#24 lore raw
Quoted 4 lines · Horst von Brand
Quoted 2 lines · Ed Cogburn
Quoted 1 line · David Lang

On Fri, 8 Jul 2005, Ed Tomlinson wrote:

Quoted 8 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

No Flame from me. One thing to remember is that Hans and friends _have_ supported R3 for years.

They let it fall into disrepair when they started work on 4.

Quoted 1 line
Quoted 1 line
Quoted 1 line

This is an undisputed fact.

Exactly.

This is FUD. Hans do you have figures on how many fixes for R3 have been added in the last year or so?

Quoted 8 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

Second
third parties have be able to add much function (like journaling) to R3 so the code must be sort of readable...

Why don't you check it? Wouldn't you much prefer if the original authors (or somebody similarly initmate with the code) did mayor surgery on it? Specially if it is something you depend on?

Quoted 8 lines
Quoted 1 line
Quoted 1 line
Quoted 1 line

With R4 they have created a new FS that is _fast_

Remains to be seen.

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

and _can_ do things
no other FS can

Mostly useless things...

Just because you do not see how to use a new feature does not mean its useless...

Quoted 14 lines
Quoted 1 line
Quoted 1 line
Quoted 1 line

- I also expect they have written cleaner code...

Better check first.

Quoted 1 line
Quoted 1 line
Quoted 1 line

Why are we fighting about adding this sort of function to the kernel?

Because the filessytem experts in the kernel development crowd (and others) have /serious/ problems with the ideas and the code?

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Yes it may not be the absolute best way to do things. How many times has tcpip be rewritten for linux? The answer is more than once.

So?

Quoted 7 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Lets put R4 in, see how it works, generalize the ideas and if we have to rewrite and rethink part of it lets do so.

Why not: Let's keep it out, fix the problems that it has and evaluate it for inclusion once the problems have been ironed out? That has been the policy for everything else as far as I can remember (and that is from nearly the beginning...)

When you are exploring new stuff its almost impossible to see the best way to do things. Once R4 (and other FSes) use these new features (or not) it will become more obvious how they should be coded. As it is not its a bit like trying to decide your new baby will be a doctor when he/she grows up - its not something you can predict.

Quoted 9 lines
Quoted 8 lines
Quoted 8 lines

remember that Hans is on record (over a year ago) arguing that R3 should not be fixed becouse R4 was replacing it.

This type of thing is one of the reasons that you see arguments that aren't 'purely code-related' becouse the kernel folks realize that _they_ will have to maintain the code over time, Hans and company will go on and develop R5 (R10, whatever) and consider R4 obsolete and stop maintaining it.

Quoted 4 lines
Quoted 2 lines

Maybe its because Hans and Co., having only a finite amount of dev time, would much prefer to spend that time on R4 rather than R3?

ext2 is still being maintained alongside ext3.

And so is R3 - just no new features are being added...

Quoted 6 lines
Quoted 3 lines

Maybe if we
were to let R4 into the kernel, it wouldn't be long after that R3 could be retired because everyone has moved to R4?

ext3 is several years old, and there are /still/ ext2 users around...

See above - this should presend NO problems.

Quoted 7 lines
Quoted 3 lines

Devs should be free to work on whatever they want, because most of them are doing this on their own time anyway, otherwise they might just decide to hack on some other OS, or a fork of Linux instead.

Nobody forces anybody to work on Linux, or even on the standard Linus kernel. It is the ReiserFS crowd who are demanding something from the Linux crowd, not the other way around.

#25 lore raw
Quoted 8 lines · Ed Tomlinson
Quoted 7 lines · Horst von Brand
Quoted 5 lines · Ed Cogburn
Quoted 4 lines · David Lang
Quoted 2 lines · Ed Tomlinson

No Flame from me. One thing to remember is that Hans and friends _have_ supported R3 for years.

Quoted 8 lines
Quoted 5 lines

They let it fall into disrepair when they started work on 4.

Quoted 1 line
Quoted 1 line
Quoted 1 line

This is an undisputed fact.

Exactly.

This is FUD. Hans do you have figures on how many fixes for R3 have been added in the last year or so?

You don't need Hans to find out how many. linux.bkbits.net is still online.

http://linux.bkbits.net:8080/linux-2.6/src/fs/reiserfs?nav=index.html|src/|src/fs

#26 lore raw
Quoted 13 lines · Alexey Dobriyan
Quoted 12 lines · Ed Tomlinson
Quoted 11 lines · Horst von Brand
Quoted 5 lines · Ed Cogburn
Quoted 4 lines · David Lang
Quoted 2 lines · Ed Tomlinson

No Flame from me. One thing to remember is that Hans and friends _have_ supported R3 for years.

They let it fall into disrepair when they started work on 4.

Quoted 1 line
Quoted 1 line
Quoted 1 line

This is an undisputed fact.

Quoted 8 lines
Quoted 4 lines
Quoted 1 line

Exactly.

This is FUD. Hans do you have figures on how many fixes for R3 have been added in the last year or so?

You don't need Hans to find out how many. linux.bkbits.net is still online.

http://linux.bkbits.net:8080/linux-2.6/src/fs/reiserfs?nav=index.html|src/|src/fs

Which seems to support the notion that namesys stopped support for reiser3 when reiser4 was started. The tracking of who submittend and who committed patches wasn't there just over a year ago so sometimes it's hard to tell who actually posted that patch if someone like Andrew or Linus committed it. But in most of the changesets on the bkbits site you can go back over 2 years and not see anything from namesys people. Nearly all of the fixes commited in the past 2-3 years are from SuSe.

Jim.

#27 lore raw
Quoted 3 lines · Jim Crilly

But in most of the changesets on the bkbits site you can go back over 2 years and not see anything from namesys people. Nearly all of the fixes commited in the past 2-3 years are from SuSe.

So, for the sake of argument, if IBM were to drop official support for JFS, we'd yank JFS out of the kernel even if there was someone else willing to support it? Why does it now *matter* who supports it, as long as its being maintained? And will we now block IBM's hypothetical JFS2 from the kernel if IBM, from the programmers up to the CEO, doesn't swear on their momma's grave that they'll continue to support JFS1, even if JFS1 is being supported by others? Jeez, this is why it doesn't take a kernel dev to see the problems here, common sense seems to be an increasingly rare ingredient in these arguments against R4. If I didn't know better, I'd think you were making this stuff up as you went along....

#28 lore raw
Quoted 5 lines · Ed Cogburn
Quoted 3 lines

But in most of the changesets on the bkbits site you can go back over 2 years and not see anything from namesys people. Nearly all of the fixes commited in the past 2-3 years are from SuSe.

Quoted 10 lines

So, for the sake of argument, if IBM were to drop official support for JFS, we'd yank JFS out of the kernel even if there was someone else willing to support it? Why does it now *matter* who supports it, as long as its being maintained? And will we now block IBM's hypothetical JFS2 from the kernel if IBM, from the programmers up to the CEO, doesn't swear on their momma's grave that they'll continue to support JFS1, even if JFS1 is being supported by others? Jeez, this is why it doesn't take a kernel dev to see the problems here, common sense seems to be an increasingly rare ingredient in these arguments against R4. If I didn't know better, I'd think you were making this stuff up as you went along....

you are completely missing the point.

the fact that the kernel group is going to have to maintain the code over the long run means that it must be acceptable to them before it gets added.

so saying that it's supported (for now) by namesys doesn't matter. if it's not something that is in a state that can be maintained over the long term by the kernel group then it can't be accepted, exactly BECOUSE nobody expects an outside orginization to maintain the code forever.

Namesys is allowed to maintain the code themselves outside the kernel for as long as they want to (and even fork the kernel if they need to make changes to it that aren't acceptable to the mainline). Namesys is asking the core kernel team to accept their code into the mainline so that it will be maintained by the core kernel team for the indefinate future. This is why it's up to Namesys to satisfy the concerns of the core team, not the other way around.

David Lang

Signature · 2 lines
There are two ways of constructing a software design. One way is to make it so simple that there are obviously no deficiencies. And the other way is to make it so complicated that there are no obvious deficiencies.
 -- C.A.R. Hoare
#29 lore raw
In reply to Ed Cogburn
Quoted 7 lines · Ed Cogburn
Quoted 3 lines · Jim Crilly

But in most of the changesets on the bkbits site you can go back over 2 years and not see anything from namesys people. Nearly all of the fixes commited in the past 2-3 years are from SuSe.

Quoted 10 lines

So, for the sake of argument, if IBM were to drop official support for JFS, we'd yank JFS out of the kernel even if there was someone else willing to support it? Why does it now *matter* who supports it, as long as its being maintained? And will we now block IBM's hypothetical JFS2 from the kernel if IBM, from the programmers up to the CEO, doesn't swear on their momma's grave that they'll continue to support JFS1, even if JFS1 is being supported by others? Jeez, this is why it doesn't take a kernel dev to see the problems here, common sense seems to be an increasingly rare ingredient in these arguments against R4. If I didn't know better, I'd think you were making this stuff up as you went along....

Someone other than Namesys maintaing the filesystem isn't the problem. XFS for instance has had a lot of work done by non-SGI employees after it's merge, but SGI doesn't release a new filesystem every 3 years with the desire to remove and replace the old one. The main problems with reiser4 have been beat to death, if you don't get it by now chances are that you won't. Having Hans and team run off to work on reiser5 6 months after inclusion is an issue, since it seems to have happened before, but it's a minor one as long as reiser4 is merged in a state where it can be manged by other people without too much trouble.

Jim.

Quoted 8 lines

-
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 Ed Tomlinson
Quoted 16 lines · Ed Tomlinson
Quoted 13 lines · Horst von Brand
Quoted 10 lines · Ed Cogburn
Quoted 7 lines · David Lang
Quoted 4 lines · Ed Tomlinson

No Flame from me. One thing to remember is that Hans and friends _have_ supported R3 for years.

Quoted 8 lines
Quoted 3 lines

They let it fall into disrepair when they started work on 4.

This is FUD. Hans do you have figures on how many fixes for R3 have been added in the last year or so?

No, but I can tell you that > 2/3 were related to features I thought should have been put in V4 instead, and were added in violation of my declared code freeze and without my consent. Naturally, those bugs were routed to the authors of those features.

There were maybe 2 bugs in the last 18 months in code not related to code freeze violations, I don't remember exactly.

There is a simple reason why Namesys no longer spends much time on bug fixes for V3. The frozen code is too stable to generate bug reports to work on, and the unfrozen code is not ours. If the unfrozen code stopped being maintained, I'd have to do something.

It seems to me that you should all hope that V4 gets to where Namesys does not spend much time maintaining it. It means we have no bug reports.

Stable branches, and development branches, and only bug fixes get added to the stable branch, it is an old paradigm, and broadly speaking it is still the right paradigm.

Guys, Horst is trolling. He has never used our code, and yet.....

#30 lore raw
In reply to Jim Crilly
Quoted 2 lines · Jim Crilly

but SGI doesn't release a new filesystem every 3 years with the desire to remove and replace the old one.

Read Han's reply to Ed T. nearby. This is why I should have followed my own original intent and not gotten back into this flamefest, as this response is the same kind of bullshit that's been repeated over and over again. My mistake, won't happen again.

#31 lore raw
Quoted 9 lines · Ed Cogburn
Quoted 2 lines · Jim Crilly

but SGI doesn't release a new filesystem every 3 years with the desire to remove and replace the old one.

Read Han's reply to Ed T. nearby. This is why I should have followed my own original intent and not gotten back into this flamefest, as this response is the same kind of bullshit that's been repeated over and over again. My mistake, won't happen again.

I did read Hans' reply, but why should I believe him? When I was skimming the csets on the bk site most of the changes looked like small fixes or code migration stuff that touched all filesystems. And I don't even know what 'features' he's saying were added that he didn't sanction. I obviously didn't read all of the log entries so it's definately possible the big stuff he's talking about was completely missed by me, but I would bet that his > 2/3 number is prett far off because every time I give reiser3 a try something eventually goes wrong and I end up back on XFS or ext3 because they're more reliable.

Jim.

#36 lore raw
In reply to Horst von Brand

Horst von Brand ([email protected]):
[on reiserfs4]

Quoted 4 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

and _can_ do things
no other FS can

Mostly useless things...

Depends on your point of view. If you define things to be useful only when POSIX requires them, then yes, reiser4 contains a lot of useless stuff.
However, it's the 'beyond POSIX'-stuff what makes reiser4 interesting.

Multistream files have been useful on other OSses for years. They might be useful on Linux too (Samba will surely like them).

The plugin architecture is very interesting. Sometimes you don't need files to be in the POSIX namespace. Why would you want to store a mysql database in files? Why not skip the overhead of the VFS and POSIX rules and just store them in a more efficient way? Maybe you can create a swapfile plugin. No need for a swapfile to be in the POSIX namespace either.
It's just a fun thing to experiment with. It's not always nescesary to let the demand create the means. Give programmers some powerful tools and wait and see what wonderful things start to evolve.

And yes, maybe in ten years time POSIX is just a subsystem in Linux. Maybe commerciale Unix vendors will start following Linux as 'the' standard instead of the other way around. Seems fun to me :-)

I think this debate will mostly boil down to 'do we want to experiment with beyond-POSIX filesystems in linux?'. Clearly we don't _need_ it now. There simply are no users. But will users come when reiser4 is merged? Nobody knows.

IMHO reiser4 should be merged and be marked as experimental. It should probably _always_ be marked as experimental, because we _know_ we're going to need some other -- more generic -- API when we decide we like the features of reiser4. The reiser4 APIs should probably be implemented as generic VFS APIs. But since we don't know yet what features we're going to use, let reiser4 be self contained. Maybe reiser5 or reiser6 will follow standard VFS-beyond-POSIX rules, with ext4 and JFS2 also implementing them.

It's just too damn hard to predict the future. IMHO better just merge reiser4 and let it be clear to everybody that reiser4 is an experiment.
As long as it doesn't affect the rest of the kernel and it's clear to the users that reiser4 is *not* going to be the standard, it's fine with me.

Signature · 1 line
Erik Hensema <[email protected]>
#32 lore raw
In reply to Ed Cogburn
Quoted 5 lines · Ed Cogburn
Quoted 3 lines · Jim Crilly

But in most of the changesets on the bkbits site you can go back over 2 years and not see anything from namesys people. Nearly all of the fixes commited in the past 2-3 years are from SuSe.

With Chris Mason's name attached? Chris wrote the journaling support for R3
and worked for SUSE for a while (he may still?). I also remember seeing quite
a few patches run though the reiser mailing list for comment...

Quoted 10 lines

So, for the sake of argument, if IBM were to drop official support for JFS, we'd yank JFS out of the kernel even if there was someone else willing to support it? Why does it now *matter* who supports it, as long as its being maintained? And will we now block IBM's hypothetical JFS2 from the kernel if IBM, from the programmers up to the CEO, doesn't swear on their momma's grave that they'll continue to support JFS1, even if JFS1 is being supported by others? Jeez, this is why it doesn't take a kernel dev to see the problems here, common sense seems to be an increasingly rare ingredient in these arguments against R4. If I didn't know better, I'd think you were making this stuff up as you went along....

#33 lore raw
Quoted 11 lines · Ed Tomlinson
Quoted 5 lines · Ed Cogburn
Quoted 3 lines · Jim Crilly

But in most of the changesets on the bkbits site you can go back over 2 years and not see anything from namesys people. Nearly all of the fixes commited in the past 2-3 years are from SuSe.

With Chris Mason's name attached? Chris wrote the journaling support for R3
and worked for SUSE for a while (he may still?). I also remember seeing quite
a few patches run though the reiser mailing list for comment...

I thought r3 was journaled from the beginning; the Namesys site credits Chris with the addition of a relocated and large journal. And yes, a good bit of the patches were from him.

Jim.

#37 lore raw
In reply to Erik Hensema
Quoted 10 lines · Erik Hensema

Horst von Brand ([email protected]):
[on reiserfs4]

Quoted 4 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

and _can_ do things
no other FS can

Mostly useless things...

Depends on your point of view. If you define things to be useful only when POSIX requires them, then yes, reiser4 contains a lot of useless stuff.

That isn't my definition.

Quoted 2 lines

However, it's the 'beyond POSIX'-stuff what makes reiser4 interesting.

I haven't seen a shred of evidence of that up to here. Just redoing in-kernel (for completely inscrutable reasons) stuff that has been confortably done in userland for many years isn't "Interesting", quite the contrary.

Quoted 1 line

Multistream files have been useful on other OSses for years.

I only have seen other OSes moving away from such stuff...

Quoted 2 lines

They
might be useful on Linux too (Samba will surely like them).

OK, if you think Windows is a good idea all around...

Quoted 1 line

The plugin architecture is very interesting.

Again: It isn't "plugins", it's "kernel configuration options redefining the filesystem layout". And that is extremely toxic: If the claims are to be believed, somebody using ReiserFS 4 could end up using filesystems as widely different as ext3 and ufs today. Both called the same. Or everybody will end up using the exact same set of "plugins", so they make no sense as configuration options. Sure, it is nice to have different versions of the same filesystem (in a way, ext3 is a version of ext2; in ext3 there are some options that where introduced later, and some of which aren't backwards-compatible), but this is not something I would want each individual user screw around with willy-nilly. So the whole "plugin" idea is very questionable to me.

Quoted 2 lines

Sometimes you don't
need files to be in the POSIX namespace.

The POSIX namespace /is/ the namespace for files.

Quoted 2 lines

Why would you want to
store a mysql database in files?

Because it is the abstraction of permanent storage that the OS gives me. Or I could write them directly on a raw block device for performance (by cutting out a middleman).

Quoted 2 lines

Why not skip the overhead of the
VFS and POSIX rules and just store them in a more efficient way?

Exactly. Cut out the filesystem.

Quoted 1 line

Maybe you can create a swapfile plugin.

The kernel manages swapping on files and devices just fine, thank you.

Quoted 2 lines

No need for a swapfile to
be in the POSIX namespace either.

And how do you handle it if it has no filename?!

Quoted 1 line

It's just a fun thing to experiment with.

Noone here is stopping you from experimenting.

Quoted 4 lines

It's not always
nescesary to let the demand create the means. Give programmers some powerful tools and wait and see what wonderful things start to evolve.

The sad truth is that if you give a random collection of people powerful tools they misuse them more often than not, creating a huge mess in the process. That is why it is so hard to design good tools.

Quoted 4 lines

And yes, maybe in ten years time POSIX is just a subsystem in Linux. Maybe commerciale Unix vendors will start following Linux as 'the' standard instead of the other way around. Seems fun to me :-)

To me too. But forcing Linux today to be as non-POSIX as possible, just so it will be prepared for 10 years in the future makes no sense, because you break it /now/.

Quoted 2 lines

I think this debate will mostly boil down to 'do we want to experiment with beyond-POSIX filesystems in linux?'.

You (and some others) clearly want to. More power to the bunch that comes up with clean semantics that can be implemented efficiently and are useful in real life (as opposed to "it would be oh-so-nice to also have $FEATURE for my pet $NICHE_CASE, feature for which I just can't be bothered considering ramifications at all"). Before going off look up "featuritis" (and consider how it all but killed off a lot of OSes, even many Unix variants, and uncountable other things too).

Quoted 2 lines

Clearly we don't _need_ it now. There simply are no users. But will users come when reiser4 is merged? Nobody knows.

Probably a tiny minority. Something like the following ReiserFS 3 has today.

Quoted 1 line

IMHO reiser4 should be merged and be marked as experimental.

IMHO ReiserFS 4 should not be merged into Linus' kernel. So what? It is not my call (nor yours).

Quoted 8 lines

It
should probably _always_ be marked as experimental, because we _know_ we're going to need some other -- more generic -- API when we decide we like the features of reiser4. The reiser4 APIs should probably be implemented as generic VFS APIs. But since we don't know yet what features we're going to use, let reiser4 be self contained. Maybe reiser5 or reiser6 will follow standard VFS-beyond-POSIX rules, with ext4 and JFS2 also implementing them.

If is is /that/ experimental, it has no place in Linus' kernel at all. It is not (and has not been for a half dozen years at least) a playground for random experiments. Sure, you can fork off a branch for fooling around, you are even wellcome to keep your laboratory synched up with the version everybody is using, if that serves your needs.

Quoted 1 line

It's just too damn hard to predict the future.

Right. Instead of gambling it will turn out just as you think (why should I give /your/ view more weight than mine? or prefer views which have shown to be erroneous over the view of people who /have/ shaped the present we are in today?), why not wait and see?

Quoted 3 lines

IMHO better just
merge reiser4 and let it be clear to everybody that reiser4 is an experiment.

IMHO much simpler just leaving experiments to experimental branches.

Quoted 1 line

As long as it doesn't affect the rest of the kernel

Impossible.

Quoted 3 lines

and it's
clear to the users that reiser4 is *not* going to be the standard, it's fine with me.

Now I'm at a complete loss... Why should it be placed in the standard kernel then?

Signature · 4 lines
Dr. Horst H. von Brand                   User #22616 counter.li.org
Departamento de Informatica                     Fono: +56 32 654431
Universidad Tecnica Federico Santa Maria              +56 32 654239
Casilla 110-V, Valparaiso, Chile                Fax:  +56 32 797513
#38 lore raw
Quoted 11 lines · Horst von Brand
Quoted 10 lines

It's not always
nescesary to let the demand create the means. Give programmers some powerful tools and wait and see what wonderful things start to evolve.

Quoted 5 lines

The sad truth is that if you give a random collection of people powerful tools they misuse them more often than not, creating a huge mess in the process. That is why it is so hard to design good tools.

We are rope makers. Our job is to make good rope. If someone might use it to hang dissidents, that does not mean we should now make the rope too inflexible to form a noose. It is important that we know our place. Our place is to help users express themselves the way they want to. It is not our job to keep them from hanging dissidents. If they hang dissidents, we should not change the way we make rope, we should
shoot them. Most of our users don't hang dissidents, to the contrary,
they do work of value to society, and need their time saved so that they can do more of it.

The users of reiser4 know a lot more than I do, and are much wiser than I am. Because I focus on a narrow little area, I am able to do something useful to help them express their greater wisdom more flexibly. I take pride in that.

The belief expressed above that powerful tools will be misused more than well used, and the dislike of power for the users it contains, is why we will never do more than talk past each other. Perhaps we should just agree to disagree.

Hans

#34 lore raw
In reply to Jim Crilly
Quoted 5 lines · Jim Crilly

I thought r3 was journaled from the beginning; the Namesys site credits Chris with the addition of a relocated and large journal. And yes, a good bit of the patches were from him.

Chris and I disagree about QA methodology, but I am deeply in debt to him for his contributions. R3 did not have a journal at first, that was his contribution, and a major part of what made reiserfs useful to real users.

Hans

#39 lore raw
In reply to Horst von Brand

Horst von Brand ([email protected]):

Quoted 6 lines
Quoted 4 lines · Erik Hensema

Horst von Brand ([email protected]):
[on reiserfs4]

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

and _can_ do things
no other FS can

Quoted 7 lines
Quoted 5 lines
Quoted 1 line

Mostly useless things...

Depends on your point of view. If you define things to be useful only when POSIX requires them, then yes, reiser4 contains a lot of useless stuff.

That isn't my definition.

Ok. It seems to be the definition of some people over here tough.

Quoted 7 lines
Quoted 2 lines

However, it's the 'beyond POSIX'-stuff what makes reiser4 interesting.

I haven't seen a shred of evidence of that up to here. Just redoing in-kernel (for completely inscrutable reasons) stuff that has been confortably done in userland for many years isn't "Interesting", quite the contrary.

Some things simply need to be done inside the kernel for atomicity, especially when mixing 'normal' posix applications with apps using extended semantics.

Quoted 3 lines
Quoted 1 line

Multistream files have been useful on other OSses for years.

I only have seen other OSes moving away from such stuff...

That would be OSX I guesss.

Quoted 4 lines
Quoted 2 lines

They
might be useful on Linux too (Samba will surely like them).

OK, if you think Windows is a good idea all around...

Windows is a fact of life, like it or not. Every Linux+Samba server replacing a Win2k3 server is a win, IMHO.

Quoted 13 lines
Quoted 1 line

The plugin architecture is very interesting.

Again: It isn't "plugins", it's "kernel configuration options redefining the filesystem layout". And that is extremely toxic: If the claims are to be believed, somebody using ReiserFS 4 could end up using filesystems as widely different as ext3 and ufs today. Both called the same. Or everybody will end up using the exact same set of "plugins", so they make no sense as configuration options. Sure, it is nice to have different versions of the same filesystem (in a way, ext3 is a version of ext2; in ext3 there are some options that where introduced later, and some of which aren't backwards-compatible), but this is not something I would want each individual user screw around with willy-nilly. So the whole "plugin" idea is very questionable to me.

Not every plugin changes the filesystem layout. Plugins can also change higher level APIs.

Changing the filesystem layout is quite dangerous indeed. Hans, is a list of required plugins stored in the superblock?

Quoted 3 lines
Quoted 1 line

need files to be in the POSIX namespace.

The POSIX namespace /is/ the namespace for files.

It doesn't need to be the namespace for _all_ files.

Quoted 6 lines
Quoted 2 lines

Why would you want to
store a mysql database in files?

Because it is the abstraction of permanent storage that the OS gives me. Or I could write them directly on a raw block device for performance (by cutting out a middleman).

Reiser4 could offer you a differend kind of abstraction. I don't know if it would be useful in the mysql case, I don't write database software.
Mysql is just a random example of software needing to store data somewhere. The data is only accessable through the mysql api. Backing up the files is useles when the server is running.

Quoted 9 lines
Quoted 3 lines

Why not skip the overhead of the
VFS and POSIX rules and just store them in a more efficient way?

Exactly. Cut out the filesystem.

Quoted 1 line

Maybe you can create a swapfile plugin.

The kernel manages swapping on files and devices just fine, thank you.

Just a random example.

Quoted 4 lines
Quoted 2 lines

No need for a swapfile to
be in the POSIX namespace either.

And how do you handle it if it has no filename?!

mkswap --kind=reiser4fsswap --operation=create

whatever.

Quoted 4 lines
Quoted 1 line

It's just a fun thing to experiment with.

Noone here is stopping you from experimenting.

Quoted 8 lines
Quoted 4 lines

It's not always
nescesary to let the demand create the means. Give programmers some powerful tools and wait and see what wonderful things start to evolve.

The sad truth is that if you give a random collection of people powerful tools they misuse them more often than not, creating a huge mess in the process. That is why it is so hard to design good tools.

Agreed. That's why you need to experiment.

Quoted 8 lines
Quoted 4 lines

And yes, maybe in ten years time POSIX is just a subsystem in Linux. Maybe commerciale Unix vendors will start following Linux as 'the' standard instead of the other way around. Seems fun to me :-)

To me too. But forcing Linux today to be as non-POSIX as possible, just so it will be prepared for 10 years in the future makes no sense, because you break it /now/.

Linux will still be posix. video4linux isn't posix either. I don't see why access to data stores should be posix all the time.

Quoted 10 lines
Quoted 2 lines

I think this debate will mostly boil down to 'do we want to experiment with beyond-POSIX filesystems in linux?'.

You (and some others) clearly want to. More power to the bunch that comes up with clean semantics that can be implemented efficiently and are useful in real life (as opposed to "it would be oh-so-nice to also have $FEATURE for my pet $NICHE_CASE, feature for which I just can't be bothered considering ramifications at all"). Before going off look up "featuritis" (and consider how it all but killed off a lot of OSes, even many Unix variants, and uncountable other things too).

Extending a filesystem beyond the standard POSIX semantics is a huge step. Not even a company as large as Microsoft can make the next step in filesystem design (WinFS). It must be done in small steps. Reiser4 is one of these small steps. Maybe the step isn't quite in the right direction, but I don't think it will be a detour either.

Quoted 5 lines
Quoted 2 lines

Clearly we don't _need_ it now. There simply are no users. But will users come when reiser4 is merged? Nobody knows.

Probably a tiny minority. Something like the following ReiserFS 3 has today.

Do you include Reiser3 users in the tiny minority? Both SuSE and Gentoo use reiser3 as their default filesystem. Both distro's are huge.
Both ext3 and reiser3 have a huge number of users. Both filesystems also have got a number of religious followers. Ignore them. Religion is no basis for a healty debate.

Quoted 4 lines
Quoted 1 line

IMHO reiser4 should be merged and be marked as experimental.

IMHO ReiserFS 4 should not be merged into Linus' kernel. So what? It is not my call (nor yours).

By having this discussion we shape our opinions. We inspire others to form their own opinion. Maybe we even inspire Linus to have an opinion on the matter. Who knows.

Quoted 8 lines
Quoted 7 lines

should probably _always_ be marked as experimental, because we _know_ we're going to need some other -- more generic -- API when we decide we like the features of reiser4. The reiser4 APIs should probably be implemented as generic VFS APIs. But since we don't know yet what features we're going to use, let reiser4 be self contained. Maybe reiser5 or reiser6 will follow standard VFS-beyond-POSIX rules, with ext4 and JFS2 also implementing them.

Quoted 5 lines

If is is /that/ experimental, it has no place in Linus' kernel at all. It is not (and has not been for a half dozen years at least) a playground for random experiments. Sure, you can fork off a branch for fooling around, you are even wellcome to keep your laboratory synched up with the version everybody is using, if that serves your needs.

I don't think it's a random experiment. Sure, it *is* an experiment. IPv6 has been an experiment for some time too.

Sometimes you simply need to experiment on a larger scale than possible simply in a private kernel fork. Why would Samba support multi-fork files when almost nobody has access to kernels supporting those files? Why would INN store its newsspool in reiser4? Why would people think about the possibilities when the standard kernel doesn't offer them?

Quoted 6 lines
Quoted 1 line

It's just too damn hard to predict the future.

Right. Instead of gambling it will turn out just as you think (why should I give /your/ view more weight than mine? or prefer views which have shown to be erroneous over the view of people who /have/ shaped the present we are in today?), why not wait and see?

Wait and see what others do? Clone their inventions? Chasing taillights? Isn't Open Source the place where innovation is supposed to happen?

Quoted 5 lines
Quoted 3 lines

IMHO better just
merge reiser4 and let it be clear to everybody that reiser4 is an experiment.

IMHO much simpler just leaving experiments to experimental branches.

The simplest option is to stop developing linux, except for new hardware drivers.

Quoted 3 lines
Quoted 1 line

As long as it doesn't affect the rest of the kernel

Impossible.

Then make sure those effects are positive. If you can say 'reiser4 got $feature wrong and I can do better' then that's a win.

Quoted 6 lines
Quoted 3 lines

and it's
clear to the users that reiser4 is *not* going to be the standard, it's fine with me.

Now I'm at a complete loss... Why should it be placed in the standard kernel then?

Scale. Small experiments are fun, large experiments are even more fun ;-)

Signature · 1 line
Erik Hensema <[email protected]>