Skip to contentSkip to conversation

MODULE_MAINTAINER

PATCH Rene Herman 13 participants 50 messages Apr 4, 2007 – Apr 28, 2007 Module support

lore
Share & tools
lore
#0 lore raw

Hi Rusty.

Can we have a MODULE_MAINTAINER to complement MODULE_AUTHOR?

Often a module grows multiple authors over time, but initial authors aren't around (or too interested) anymore. And sometimes, if someone maintaining a driver is just doing minor peripheral updates to keep it compiling, the maintainer might not even _be_ an author really...

Rene.

include/linux/module.h +3 -0
index 10f771a..c06f76c 100644
@@ -128,6 +128,9 @@ #define MODULE_LICENSE(_license) MODULE_
 /* Author, ideally of form NAME <EMAIL>[, NAME <EMAIL>]*[ and NAME <EMAIL>] */
 #define MODULE_AUTHOR(_author) MODULE_INFO(author, _author)
   
+/* Maintainer, ideally of form NAME <EMAIL> */
+#define MODULE_MAINTAINER(_maintainer) MODULE_INFO(maintainer, _maintainer)
+
 /* What your module does. */
 #define MODULE_DESCRIPTION(_description) MODULE_INFO(description, _description)
 
#1 lore raw
Quoted 1 line · Rene Herman

Can we have a MODULE_MAINTAINER to complement MODULE_AUTHOR?

And here's the accompanying patch to the module-init-tools-3.3-pre1 as found on http://kernel.org/pub/linux/utils/kernel/module-init-tools/

Rene.

modinfo.c renamed from modinfo.c.orig +3 -1
@@ -213,6 +213,7 @@
 static struct option options[] =
 {
 	{"author", 0, 0, 'a'},
+	{"maintainer", 0, 0, 'm'},
 	{"description", 0, 0, 'd'},
 	{"license", 0, 0, 'l'},
 	{"parameters", 0, 0, 'p'},
@@ -347,9 +348,10 @@
 	else
 		abort();
 
-	while ((opt = getopt_long(argc,argv,"adlpVhn0F:",options,NULL)) >= 0){
+	while ((opt = getopt_long(argc,argv,"amdlpVhn0F:",options,NULL)) >= 0){
 		switch (opt) {
 		case 'a': field = "author"; break;
+		case 'm': field = "maintainer"; break;	  
 		case 'd': field = "description"; break;
 		case 'l': field = "license"; break;
 		case 'p': field = "parm"; break;
#2 lore raw
In reply to Rene Herman
Quoted 3 lines · Rene Herman

Hi Rusty.

Can we have a MODULE_MAINTAINER to complement MODULE_AUTHOR?

#define MODULE_MAINTAINER(x) MODULE_AUTHOR(x), please. MODULE_AUTHOR really has meant maintainer in practice for ages, and it's the only actually relevant for users information we should store.

#3 lore raw
Quoted 7 lines · Christoph Hellwig
Quoted 1 line · Rene Herman

Can we have a MODULE_MAINTAINER to complement MODULE_AUTHOR?

#define MODULE_MAINTAINER(x) MODULE_AUTHOR(x), please. MODULE_AUTHOR really has meant maintainer in practice for ages, and it's the only actually relevant for users information we should store.

Given modules with multiple authors, current and non-current, I believe having "modinfo -m" tell the user whom to contact is an avantage.

Rene.

#4 lore raw
Quoted 4 lines · Rene Herman
Quoted 1 line · Christoph Hellwig

On Wed, Apr 04, 2007 at 01:26:04PM +0200, Rene Herman wrote:

Quoted 8 lines
Quoted 5 lines
Quoted 1 line

Can we have a MODULE_MAINTAINER to complement MODULE_AUTHOR?

#define MODULE_MAINTAINER(x) MODULE_AUTHOR(x), please. MODULE_AUTHOR really has meant maintainer in practice for ages, and it's the only actually relevant for users information we should store.

Given modules with multiple authors, current and non-current, I believe having "modinfo -m" tell the user whom to contact is an avantage.

Much bigger problems are:
- Who will maintain this information properly?
- What about modules that are maintained implicitely by the subsystem maintainer?

And often a user can't be expected to locate the source of a problem, or it might not be in a driver but in a subsystem.

For vendor kernels, the user should contact the vendor.
For ftp.kernel.org kernels, I don't see any better solution than telling people to report problems to linux-kernel or the kernel Bugzilla and routing them further from here.

Quoted 1 line

Rene.

cu
Adrian

Signature · 4 lines
       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed
#5 lore raw
Quoted 9 lines · Adrian Bunk
Quoted 2 lines · Rene Herman

Given modules with multiple authors, current and non-current, I believe having "modinfo -m" tell the user whom to contact is an avantage.

Much bigger problems are:
- Who will maintain this information properly?
- What about modules that are maintained implicitely by the subsystem maintainer?

Quoted 7 lines

And often a user can't be expected to locate the source of a problem, or it might not be in a driver but in a subsystem.

For vendor kernels, the user should contact the vendor.
For ftp.kernel.org kernels, I don't see any better solution than telling people to report problems to linux-kernel or the kernel Bugzilla and routing them further from here.

Generelly it has to be kept in mind that there are different contacts for different purposes:
- usage problems --> get in touch with the _support_
- bug reports --> get in touch with _maintainers_
- development --> get in touch with maintainers/ kernel hackers/ copyright holders...

The Amiga keyboard had a [Help] key, but this is not how it works.

Signature · 3 lines
Stefan Richter
-=====-=-=== -=-- --=--
http://arcgraph.de/sr/
#6 lore raw
Quoted 18 lines · Stefan Richter
Quoted 9 lines · Adrian Bunk
Quoted 2 lines · Rene Herman

Given modules with multiple authors, current and non-current, I believe having "modinfo -m" tell the user whom to contact is an avantage.

Much bigger problems are:
- Who will maintain this information properly?
- What about modules that are maintained implicitely by the subsystem maintainer?

Quoted 7 lines

And often a user can't be expected to locate the source of a problem, or it might not be in a driver but in a subsystem.

For vendor kernels, the user should contact the vendor.
For ftp.kernel.org kernels, I don't see any better solution than telling people to report problems to linux-kernel or the kernel Bugzilla and routing them further from here.

Quoted 4 lines

Generelly it has to be kept in mind that there are different contacts for different purposes:
- usage problems --> get in touch with the _support_
- bug reports --> get in touch with _maintainers_

- bug reports against vendor kernels -> get in touch with the _vendor_ (the vendor might ship a heavily patched driver in an ancient kernel)

Quoted 6 lines

- development --> get in touch with maintainers/ kernel hackers/ copyright holders...

The Amiga keyboard had a [Help] key, but this is not how it works.

Stefan Richter

cu
Adrian

Signature · 4 lines
       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed
#7 lore raw
Quoted 6 lines · Adrian Bunk
Quoted 2 lines · Stefan Richter

- usage problems --> get in touch with the _support_
- bug reports --> get in touch with _maintainers_

- bug reports against vendor kernels -> get in touch with the _vendor_ (the vendor might ship a heavily patched driver in an ancient kernel)

Yes, of course. I implicitly meant the maintainer of the program as distributed, which in case of distro kernels are the distributor's kernel maintainers... :-)

Also, "get in touch" almost never means "send personal e-mail", but rather to use bugzillas, mailing lists, discussion forums etc.

Signature · 3 lines
Stefan Richter
-=====-=-=== -=-- --=--
http://arcgraph.de/sr/
#8 lore raw
In reply to Christoph Hellwig

Hi Christoph,

Quoted 5 lines
Quoted 1 line

Can we have a MODULE_MAINTAINER to complement MODULE_AUTHOR?

#define MODULE_MAINTAINER(x) MODULE_AUTHOR(x), please. MODULE_AUTHOR really has meant maintainer in practice for ages, and it's the only actually relevant for users information we should store.

I agree. The actual author information belong into the files copyright header. The MODULE_AUTHOR should only show the current maintainer and its email address. However this might need some cleanup. Maybe it is time that we sync the MAINTAINERS file with the MODULE_AUTHOR tag.

Regards

Marcel

#9 lore raw
Quoted 4 lines · Marcel Holtmann

Hi Christoph,

Quoted 2 lines
Quoted 1 line

Can we have a MODULE_MAINTAINER to complement MODULE_AUTHOR?

Quoted 8 lines
Quoted 3 lines

#define MODULE_MAINTAINER(x) MODULE_AUTHOR(x), please. MODULE_AUTHOR really has meant maintainer in practice for ages, and it's the only actually relevant for users information we should store.

I agree. The actual author information belong into the files copyright header. The MODULE_AUTHOR should only show the current maintainer and its email address. However this might need some cleanup. Maybe it is time that we sync the MAINTAINERS file with the MODULE_AUTHOR tag.

Or even remove the MODULE_AUTHOR tag?

Even if you would spend the time for syncing the > 1600 MODULE_AUTHOR tags today, I don't see this happen on an ongoing basis in the future.

Quoted 3 lines

Regards

Marcel

cu
Adrian

Signature · 4 lines
       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed
#10 lore raw
Quoted 10 lines · Adrian Bunk
Quoted 9 lines
Quoted 3 lines

#define MODULE_MAINTAINER(x) MODULE_AUTHOR(x), please. MODULE_AUTHOR really has meant maintainer in practice for ages, and it's the only actually relevant for users information we
should store.

I agree. The actual author information belong into the files copyright header. The MODULE_AUTHOR should only show the current maintainer and its email address. However this might need some cleanup. Maybe it is time that we sync the MAINTAINERS file with the MODULE_AUTHOR tag.

Quoted 5 lines

Or even remove the MODULE_AUTHOR tag?

Even if you would spend the time for syncing the > 1600 MODULE_AUTHOR tags today, I don't see this happen on an ongoing basis in the future.

Given that people seem to agree that authorship information has no place in the binary, that might actually be best.

The problem I'm trying to solve is letting people know whom to contact. Having multiple authors in the MODULES_AUTHOR tag and maintainers who might not be authors are two reasons the current tag doesn't work.

Keeping copyright holders in a copyright header and maintainers in the MAINTAINERS file would at least avoid the confusing message the tag sends. As you say in a previous message, users should mostly in fact be contacting their vendor if using distro kernels or, if they're using kermel.org kernels, linux-kernel and can then find the information in the MAINTAINERS file.

So, MODULE_AUTHOR be gone?

Rene.

#11 lore raw
Quoted 2 lines

Given that people seem to agree that authorship information has no place in the binary, that might actually be best.

Authorship information is very useful in the binary, especially when you have to get lawyers involved in explaining things to people. Company business and management people understand

"load it into a binary editor, search for Alan Cox, now call your lawyer"

trying to prove a given source is a given binary is much much harder.

Quoted 1 line

So, MODULE_AUTHOR be gone?

Not if I have anything to do with it. Putting maintainer in is not a bad idea but that assumes it gets maintained, the beauty of _AUTHOR is that it's generally right and stays that way or approximately so.

Alan

#12 lore raw

Hi Alan,

Quoted 9 lines
Quoted 2 lines

Given that people seem to agree that authorship information has no place in the binary, that might actually be best.

Authorship information is very useful in the binary, especially when you have to get lawyers involved in explaining things to people. Company business and management people understand

"load it into a binary editor, search for Alan Cox, now call your lawyer"

Quoted 7 lines

trying to prove a given source is a given binary is much much harder.

Quoted 1 line

So, MODULE_AUTHOR be gone?

Not if I have anything to do with it. Putting maintainer in is not a bad idea but that assumes it gets maintained, the beauty of _AUTHOR is that it's generally right and stays that way or approximately so.

removing the MODULE_AUTHOR is a bad idea. I really like to call modinfo and see who maintains a specific driver.

Regards

Marcel

#13 lore raw
In reply to Alan Cox
Quoted 4 lines · Alan Cox
Quoted 1 line

Given that people seem to agree that authorship information has no place in the binary, that might actually be best.

Authorship information is very useful in the binary, especially when you have to get lawyers involved in explaining things to people.

Okay.

Quoted 5 lines
Quoted 1 line

So, MODULE_AUTHOR be gone?

Not if I have anything to do with it. Putting maintainer in is not a bad idea but that assumes it gets maintained, the beauty of _AUTHOR is that it's generally right and stays that way or approximately so.

Case in point; someone is working with me in private on a new "mitsumi" legacy CD-ROM driver. He's authoring the actual driver and upto now I've just been doing some peripheral module infrastructure work. Given that I have the hardware to test the thing, I'll be the maintainer though.

Adding myself as a MODULE_AUTHOR would be largely incorrect and adding myself as the _only_ MODULE_AUTHOR would be so factually incorrect I wouldn't, even if only from a credits point of view. Yet I do want to make sure people contact me, and not the MODULE_AUTHOR (which will happen no matter the MAINTAINERS file).

Other cases-in-point; I've lately been rummaging through sound/isa a bit. Nothing much copyrightable again but especially in those situations where (some of the) original authors are no longer active, I do again want people to contact me about them if needed. And all the "which one of the three people listed here is maintaining this" is yet another.

MODULE_AUTHOR may be approximately right but especially with old drivers it also has little relation with who's maintaining the thing.

If MODULE_AUTHOR stays, can I just have MODULE_MAINTAINER please? It doesn't need to be added to drivers directly, it can just grow (and being inside the code, I suppose it'll likely stay up to date better than the MAINTAINERS file).

Rene.

#14 lore raw

At Wed, 04 Apr 2007 18:38:42 +0200,

Quoted 6 lines · Rene Herman

Other cases-in-point; I've lately been rummaging through sound/isa a bit. Nothing much copyrightable again but especially in those situations where (some of the) original authors are no longer active, I do again want people to contact me about them if needed. And all the "which one of the three people listed here is maintaining this" is yet another.

Practically, the information can go through subsystem maintainers to specific person (you) in such a case, as the tree is still more or less maintained by subsystem maintainers.

But in general, it would be nice to have an easy way to find a maintainer (not author) from a module binary, and I agree MODULE_MAINTAINER can work well for such a purpose. It's no mandatory field, but could be some help.

Takashi

#15 lore raw
Quoted 9 lines · Takashi Iwai

At Wed, 04 Apr 2007 18:38:42 +0200,

Quoted 6 lines · Rene Herman

Other cases-in-point; I've lately been rummaging through sound/isa a bit. Nothing much copyrightable again but especially in those situations where (some of the) original authors are no longer active, I do again want people to contact me about them if needed. And all the "which one of the three people listed here is maintaining this" is yet another.

Quoted 8 lines

Practically, the information can go through subsystem maintainers to specific person (you) in such a case, as the tree is still more or less maintained by subsystem maintainers.

But in general, it would be nice to have an easy way to find a maintainer (not author) from a module binary, and I agree MODULE_MAINTAINER can work well for such a purpose. It's no mandatory field, but could be some help.

Yes, it would be nice.

But would this information always be kept up-to-date for the whole tree? I don't see this happen.

Quoted 1 line

Takashi

cu
Adrian

Signature · 4 lines
       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed
#16 lore raw
Quoted 11 lines · Adrian Bunk
Quoted 4 lines · Takashi Iwai

But in general, it would be nice to have an easy way to find a maintainer (not author) from a module binary, and I agree MODULE_MAINTAINER can work well for such a purpose. It's no mandatory field, but could be some help.

Yes, it would be nice.

But would this information always be kept up-to-date for the whole tree? I don't see this happen.

I believe it would largely stay up to date yes. The tag wouldn't be mandatory and adding oneself as a MODULE_MAINTAINER would specifically be saying "yes, I want to look after this thing". If someone then no longer wants to, getting rid of the tag is a matter of deleting one line. I wouldn't be worse than MAINTAINERS, and being inline, I expect it to be better...

Rene.

#17 lore raw
Quoted 14 lines · Rene Herman
Quoted 11 lines · Adrian Bunk
Quoted 4 lines · Takashi Iwai

But in general, it would be nice to have an easy way to find a maintainer (not author) from a module binary, and I agree MODULE_MAINTAINER can work well for such a purpose. It's no mandatory field, but could be some help.

Yes, it would be nice.

But would this information always be kept up-to-date for the whole tree? I don't see this happen.

Quoted 6 lines

I believe it would largely stay up to date yes. The tag wouldn't be mandatory and adding oneself as a MODULE_MAINTAINER would specifically be saying "yes, I want to look after this thing". If someone then no longer wants to, getting rid of the tag is a matter of deleting one line. I wouldn't be worse than MAINTAINERS, and being inline, I expect it to be better...

- often module maintainers disappear suddenly or slowly without any clear statement that maintainership ended
- most drivers are not maintained on a per-driver but on a per-subsystem basis

Consider that we don't even manage to keep MAINTAINERS correct, and consider that we are talking about more than 2800 modules.

Realistically, users should report problems with vendor kernels to the vendor and problems with ftp.kernel.org kernels to either linux-kernel or the kernel Bugzilla, and forwarding issues to the responsible people (if any) should be done there [1].

Quoted 1 line

Rene.

cu
Adrian

[1] Andrew is doing this

Signature · 4 lines
       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed
#18 lore raw
Quoted 11 lines · Adrian Bunk

Realistically, users should report problems with vendor kernels to the vendor and problems with ftp.kernel.org kernels to either linux-kernel or the kernel Bugzilla, and forwarding issues to the responsible people (if any) should be done there [1].

Quoted 1 line

Rene.

cu
Adrian

[1] Andrew is doing this

At bugzilla.kernel.org, driver maintainers can also watch the respective subsystem maintainer's address or subsystem's meta address to get notified when a bug is being filed under a subsystem.

Signature · 3 lines
Stefan Richter
-=====-=-=== -=-- --=-=
http://arcgraph.de/sr/
#19 lore raw
In reply to Rene Herman

On 04/04/2007 06:38 PM, Rene Herman wrote:

Rusty?

Quoted 17 lines · Alan Cox
Quoted 5 lines
Quoted 2 lines

Given that people seem to agree that authorship information has no place in the binary, that might actually be best.

Authorship information is very useful in the binary, especially when you have to get lawyers involved in explaining things to people.

Okay.

Quoted 5 lines
Quoted 1 line

So, MODULE_AUTHOR be gone?

Not if I have anything to do with it. Putting maintainer in is not a bad idea but that assumes it gets maintained, the beauty of _AUTHOR is that it's generally right and stays that way or approximately so.

Case in point; someone is working with me in private on a new "mitsumi" legacy CD-ROM driver. He's authoring the actual driver and upto now I've just been doing some peripheral module infrastructure work. Given that I have the hardware to test the thing, I'll be the maintainer though.

Quoted 7 lines

Adding myself as a MODULE_AUTHOR would be largely incorrect and adding myself as the _only_ MODULE_AUTHOR would be so factually incorrect I wouldn't, even if only from a credits point of view. Yet I do want to make sure people contact me, and not the MODULE_AUTHOR (which will happen no matter the MAINTAINERS file).

Other cases-in-point; I've lately been rummaging through sound/isa a bit. Nothing much copyrightable again but especially in those situations where (some of the) original authors are no longer active, I do again want people to contact me about them if needed. And all the "which one of the three people listed here is maintaining this" is yet another.

MODULE_AUTHOR may be approximately right but especially with old drivers it also has little relation with who's maintaining the thing.

If MODULE_AUTHOR stays, can I just have MODULE_MAINTAINER please? It doesn't need to be added to drivers directly, it can just grow (and being inside the code, I suppose it'll likely stay up to date better than the MAINTAINERS file).

Rene.

#20 lore raw
Quoted 3 lines · Rene Herman

On 04/04/2007 06:38 PM, Rene Herman wrote:

Rusty?

Valid points have been made on both sides. I suggest:

#define MODULE_MAINTAINER(_maintainer) \
MODULE_AUTHOR("(Maintained by) "_maintainer)

Cheers,
Rusty.

#21 lore raw
Quoted 9 lines · Rusty Russell
Quoted 3 lines · Rene Herman

On 04/04/2007 06:38 PM, Rene Herman wrote:

Rusty?

Valid points have been made on both sides. I suggest:

#define MODULE_MAINTAINER(_maintainer) \
MODULE_AUTHOR("(Maintained by) "_maintainer)

why bring MODULE_AUTHOR into it? just define it in terms of MODULE_INFO:

#define MODULE_MAINTAINER(_m) MODULE_INFO(_m, "(Maintained by)" \
maintainer)

technically, the maintainer is not the same as the author so why confuse the issue with an extra unnecessary macro expansion?

rday

Signature · 7 lines
========================================================================
Robert P. J. Day
Linux Consulting, Training and Annoying Kernel Pedantry
Waterloo, Ontario, CANADA

http://fsdev.net/wiki/index.php?title=Main_Page
========================================================================
#22 lore raw
Quoted 7 lines · Robert P. J. Day
Quoted 4 lines · Rusty Russell

Valid points have been made on both sides. I suggest:

#define MODULE_MAINTAINER(_maintainer) \
MODULE_AUTHOR("(Maintained by) "_maintainer)

Quoted 8 lines

why bring MODULE_AUTHOR into it? just define it in terms of MODULE_INFO:

#define MODULE_MAINTAINER(_m) MODULE_INFO(_m, "(Maintained by)" \
maintainer)

technically, the maintainer is not the same as the author so why confuse the issue with an extra unnecessary macro expansion?

Swap arguments to MODULE_INFO, but yes, other than the extra "(Maintained by)" that's what was originally submitted:

http://lkml.org/lkml/2007/4/4/170

If you're going to be using a "maintainer" tag anyway as both that and yours above does, the "(Maintained by)" becomes superfluous, so we're back at the original.

I must say I'm not particularly sure either why reusing MODULE_AUTHOR would be better if MODULE_AUTHOR also remains (as Alan Cox pointed out might be desirable for legal reasons if nothing else). As fas as I'm aware, the other trivial patch I posted to init-module-tools:

http://lkml.org/lkml/2007/4/4/171

is all that's needed to make it useful.

But, I also really only care about being able to add MODULE_MAINTAINER() to some drivers that have outlived their authors and from that viewpoint there is no difference, so if Rusty feels this is better, so be it.

Andrew, mind if I submit this to you?

===
Provide MODULE_MAINTAINER() as a convenient place to stick a name and email address both for drivers having multiple (current and non-current) authors and for when someone who wants to maintain a driver isn't so much an author.

===

Rene.

include/linux/module.h +4 -0
index 10f771a..3c54774 100644
@@ -128,6 +128,10 @@ extern struct module __this_module;
 /* Author, ideally of form NAME <EMAIL>[, NAME <EMAIL>]*[ and NAME <EMAIL>] */
 #define MODULE_AUTHOR(_author) MODULE_INFO(author, _author)
   
+/* Maintainer, ideally of form NAME <EMAIL> */
+#define MODULE_MAINTAINER(_maintainer) \
+	MODULE_AUTHOR("(Maintained by) "_maintainer)
+
 /* What your module does. */
 #define MODULE_DESCRIPTION(_description) MODULE_INFO(description, _description)
 
#23 lore raw
Quoted 17 lines · On Mon, 23 Apr 2007 14:32:36 +0200 Rene Herman <[email protected]> wrote:

Provide MODULE_MAINTAINER() as a convenient place to stick a name and email address both for drivers having multiple (current and non-current) authors and for when someone who wants to maintain a driver isn't so much an author.

===

Rene.

[module_maintainer2.diff text/plain (604B)]

include/linux/module.h +0 -0
index 10f771a..3c54774 100644

include/linux/module.h extern struct module __this_module; · line 134

Quoted 7 lines
include/linux/module.h +4 -0
@@ -128,6 +128,10 @@ extern struct module __this_module;
 /* Author, ideally of form NAME <EMAIL>[, NAME <EMAIL>]*[ and NAME <EMAIL>] */
 #define MODULE_AUTHOR(_author) MODULE_INFO(author, _author)
 
+/* Maintainer, ideally of form NAME <EMAIL> */
+#define MODULE_MAINTAINER(_maintainer) \
+	MODULE_AUTHOR("(Maintained by) "_maintainer)
+

I'm not sure we want to do this - that's what ./MAINTAINERS is for and we end up having to maintain the same info in two places.

I actually use git-whatchanged if I'm unsure who to blame^Wask for help on a particular piece of code.

An easy way of doing this is to go to
http://git.kernel.org/gitweb.cgi?p=linux/kernel/git/torvalds/linux-2.6.git;a=tree
then drill down to the file and hit the "history" link. That will tell you who is *really* doing work on the particular code.

#24 lore raw
Quoted 6 lines · Andrew Morton
Quoted 1 line · On Mon, 23 Apr 2007 14:32:36 +0200 Rene Herman <[email protected]> wrote:

Provide MODULE_MAINTAINER() as a convenient place to stick a name and email

I'm not sure we want to do this - that's what ./MAINTAINERS is for and we end up having to maintain the same info in two places.

I don't think MAINTAINERS should cover most modules. But I don't really mind.

Rusty.

#25 lore raw
In reply to Andrew Morton
Quoted 7 lines · Andrew Morton
Quoted 4 lines · On Mon, 23 Apr 2007 14:32:36 +0200 Rene Herman <[email protected]> wrote:

Provide MODULE_MAINTAINER() as a convenient place to stick a name and email address both for drivers having multiple (current and non-current) authors and for when someone who wants to maintain a driver isn't so much an author.

[ snip ]

Quoted 2 lines

I'm not sure we want to do this - that's what ./MAINTAINERS is for and we end up having to maintain the same info in two places.

joe@user:~$ less ./MAINTAINERS
./MAINTAINERS: No such file or directory

MAINTAINERS is a developers thing, not users, yet a maintainer is someone who other than by developers wants to be contacted by users of a particular driver. Right now, a module exports a set of name and email addresses through the MODULE_AUTHOR tag but given multiple current and non-current authors, completely or largely orphaned drivers (I have a lot of junk PC hardware so I come across those relatively often) and people who might be interested in taking care of a driver but who do not consider themselves an author for (upto now) having done a s/, struct pt_regs// on it, that tag only confuses the issue of whom to contact.

And it in fact even does so when Joe does know about a MAINTAINERS file and does happen to have a kernel source tree lying around somewhere. With one set of addresses displayed prominently inside the sourcecode of the very driver and another one of in a MAINTAINERS file, the first one wins. Joe would have to be very new to Linux to trust something in the tree that's not actually compiled over something that is.

As the first response in this thread Cristoph Hellwig stated that MODULE_AUTHOR serves no purpose other than what MODULE_MAINTAINER would be serving. Others agreed and Adrian Bunk suggested deleting MODULE_AUTHOR outright.

That would actually also serve my purposes; if there's no MODULE_AUTHOR confusing the issue, I don't so much need a MODULE_MAINTAINER to fix it again. I believe having "modinfo" (optionally!) display a contact address for a driver might be a user advantage, but with all the wrong addresses gone, I don't really care deeply; MODULE_AUTHOR doesn't serve the purpose today and with it gone the user at least knows he needs to look elsewhere. MODULE_AUTHOR is also a credits issue but the information can be transferred to copyright headers. It would obviously also fix any possible maintenance issues.

Alan Cox believes that having author information embedded in the module serves a legal purpose though and objects to removal.

So, if MODULE_AUTHOR is going to stay I'd like to have MODULE_MAINTAINER to fix the message it sends. Some of it could be fixed by just deleting the email addresses from the MODULE_AUTHOR tag but that's probably not a good solution _either_. Those possible legal purposes are then the only conceivable use of the tag meaning that it should either be deleted outright if people don't agree on the legal angle or should retain the address as a contact point for legal issues/questions if people do.

The purpose I want MODULE_MAINTAINER for would not introduce any significant maintenance issues. It's not a "whole tree or nothing" thing. I want it for a few ISA crap drivers that have outlived their authors but list their names and email addresses in the source and binary through the MODULE_AUTHOR tag;
simply going around deleting MODULE_AUTHOR tags is not something us random
kernelnewbie wannabee driver hackers can do. It's something "the community" could do but if noone is going to, we don't have a good way to override the information from MODULE_AUTHOR.

Rene.

#26 lore raw
Quoted 49 lines · Rene Herman
Quoted 6 lines · Andrew Morton
Quoted 4 lines · On Mon, 23 Apr 2007 14:32:36 +0200 Rene Herman <[email protected]> wrote:

Provide MODULE_MAINTAINER() as a convenient place to stick a name and email address both for drivers having multiple (current and non-current) authors and for when someone who wants to maintain a driver isn't so much an author.

[ snip ]

Quoted 2 lines

I'm not sure we want to do this - that's what ./MAINTAINERS is for and we end up having to maintain the same info in two places.

joe@user:~$ less ./MAINTAINERS
./MAINTAINERS: No such file or directory

MAINTAINERS is a developers thing, not users, yet a maintainer is someone who other than by developers wants to be contacted by users of a particular driver. Right now, a module exports a set of name and email addresses through the MODULE_AUTHOR tag but given multiple current and non-current authors, completely or largely orphaned drivers (I have a lot of junk PC hardware so I come across those relatively often) and people who might be interested in taking care of a driver but who do not consider themselves an author for (upto now) having done a s/, struct pt_regs// on it, that tag only confuses the issue of whom to contact.

And it in fact even does so when Joe does know about a MAINTAINERS file and does happen to have a kernel source tree lying around somewhere. With one set of addresses displayed prominently inside the sourcecode of the very driver and another one of in a MAINTAINERS file, the first one wins. Joe would have to be very new to Linux to trust something in the tree that's not actually compiled over something that is.

As the first response in this thread Cristoph Hellwig stated that MODULE_AUTHOR serves no purpose other than what MODULE_MAINTAINER would be serving. Others agreed and Adrian Bunk suggested deleting MODULE_AUTHOR outright.

That would actually also serve my purposes; if there's no MODULE_AUTHOR confusing the issue, I don't so much need a MODULE_MAINTAINER to fix it again. I believe having "modinfo" (optionally!) display a contact address for a driver might be a user advantage, but with all the wrong addresses gone, I don't really care deeply; MODULE_AUTHOR doesn't serve the purpose today and with it gone the user at least knows he needs to look elsewhere. MODULE_AUTHOR is also a credits issue but the information can be transferred to copyright headers. It would obviously also fix any possible maintenance issues.

Quoted 3 lines

Alan Cox believes that having author information embedded in the module serves a legal purpose though and objects to removal.
...

Let me try to summarize the points:
- you think MODULE_AUTHOR without MODULE_MAINTAINER confuses users
- Alan wants MODULE_AUTHOR to stay for easing showing authorship of code to people
- I (and others) think MODULE_MAINTAINER wouldn't make sense

Is there any good solution for this?
E.g. modinfo could be changed to no longer defaulting to show the author?

Quoted 1 line

Rene.

cu
Adrian

Signature · 4 lines
       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed
#27 lore raw
Quoted 2 lines · Adrian Bunk

Let me try to summarize the points:
- you think MODULE_AUTHOR without MODULE_MAINTAINER confuses users

Yes, and the frustration of composing lengthy emails to bouncing (or worse still, silent) email adresses is severe if you just decided to for once not shrug about it but submit a bugreport. If it happens to an address from MAINTAINERS (and an LKML savvy user) the address can at least be deleted but you can't delete a MODULE_AUTHOR tag.

Without something like a MODULE_MAINTAINER I have no way of saying "ignore those MODULE_AUTHORs please, I want the bugreport".

Quoted 3 lines

- Alan wants MODULE_AUTHOR to stay for easing showing authorship of code to people
- I (and others) think MODULE_MAINTAINER wouldn't make sense

Had a few Ays as well. But note, since you suggested removing MODULE_AUTHOR I actually agree that made more sense. It's just that iff that can't happen, I feel I need a place to override MODULE_AUTHOR for the use that I'm interested in and which is acting as the contact for a driver.

Quoted 2 lines

Is there any good solution for this?
E.g. modinfo could be changed to no longer defaulting to show the author?

Tie Alan to his chair and take away his keyboard while we submit patches removing MODULE_AUTHOR? Or just apply a trivial two line, optional, non mandatory, patch introducing a MODULE_MAINTAINER? You pick... :-)

Rene.

#28 lore raw
Quoted 3 lines

Tie Alan to his chair and take away his keyboard while we submit patches removing MODULE_AUTHOR? Or just apply a trivial two line, optional, non mandatory, patch introducing a MODULE_MAINTAINER? You pick... :-)

MODULE_AUTHOR is extremely important for licensing enforcement. Removing it should not be an option.

Alan

#29 lore raw
Quoted 4 lines · Alan Cox
Quoted 1 line

Tie Alan to his chair and take away his keyboard while we submit patches removing MODULE_AUTHOR? Or just apply a trivial two line, optional, non mandatory, patch introducing a MODULE_MAINTAINER? You pick... :-)

MODULE_AUTHOR is extremely important for licensing enforcement. Removing it should not be an option.

Yes, the "choice" was obviously in jest. If you say MODULE_AUTHOR must stay, so must it. As said, that just leaves me wanting a way to override it as a contact.

Rene.

#30 lore raw
In reply to Adrian Bunk
Quoted 50 lines · On Thu, 26 Apr 2007 15:54:26 +0200 Adrian Bunk wrote:
Quoted 49 lines · Rene Herman
Quoted 6 lines · Andrew Morton
Quoted 4 lines · On Mon, 23 Apr 2007 14:32:36 +0200 Rene Herman <[email protected]> wrote:

Provide MODULE_MAINTAINER() as a convenient place to stick a name and email address both for drivers having multiple (current and non-current) authors and for when someone who wants to maintain a driver isn't so much an author.

[ snip ]

Quoted 2 lines

I'm not sure we want to do this - that's what ./MAINTAINERS is for and we end up having to maintain the same info in two places.

joe@user:~$ less ./MAINTAINERS
./MAINTAINERS: No such file or directory

MAINTAINERS is a developers thing, not users, yet a maintainer is someone who other than by developers wants to be contacted by users of a particular driver. Right now, a module exports a set of name and email addresses through the MODULE_AUTHOR tag but given multiple current and non-current authors, completely or largely orphaned drivers (I have a lot of junk PC hardware so I come across those relatively often) and people who might be interested in taking care of a driver but who do not consider themselves an author for (upto now) having done a s/, struct pt_regs// on it, that tag only confuses the issue of whom to contact.

And it in fact even does so when Joe does know about a MAINTAINERS file and does happen to have a kernel source tree lying around somewhere. With one set of addresses displayed prominently inside the sourcecode of the very driver and another one of in a MAINTAINERS file, the first one wins. Joe would have to be very new to Linux to trust something in the tree that's not actually compiled over something that is.

As the first response in this thread Cristoph Hellwig stated that MODULE_AUTHOR serves no purpose other than what MODULE_MAINTAINER would be serving. Others agreed and Adrian Bunk suggested deleting MODULE_AUTHOR outright.

That would actually also serve my purposes; if there's no MODULE_AUTHOR confusing the issue, I don't so much need a MODULE_MAINTAINER to fix it again. I believe having "modinfo" (optionally!) display a contact address for a driver might be a user advantage, but with all the wrong addresses gone, I don't really care deeply; MODULE_AUTHOR doesn't serve the purpose today and with it gone the user at least knows he needs to look elsewhere. MODULE_AUTHOR is also a credits issue but the information can be transferred to copyright headers. It would obviously also fix any possible maintenance issues.

Quoted 2 lines
Quoted 2 lines

Alan Cox believes that having author information embedded in the module serves a legal purpose though and objects to removal.

Wouldn't a /* comment */ satisfy AUTHOR needs?

It gives deserved attribution and should serve legal purpose just as well as a macro does (IANAL!).

IMO we want MAINTAINER info in the macro and in modinfo, so I'm for removing MODULE_AUTHOR and just having MAINTAINER.

Quoted 10 lines
Quoted 1 line

...

Let me try to summarize the points:
- you think MODULE_AUTHOR without MODULE_MAINTAINER confuses users
- Alan wants MODULE_AUTHOR to stay for easing showing authorship of code to people
- I (and others) think MODULE_MAINTAINER wouldn't make sense

Is there any good solution for this?
E.g. modinfo could be changed to no longer defaulting to show the author?

---
~Randy
*** Remember to use Documentation/SubmitChecklist when testing your code ***

#31 lore raw
Quoted 52 lines · Randy Dunlap
Quoted 50 lines · On Thu, 26 Apr 2007 15:54:26 +0200 Adrian Bunk wrote:
Quoted 49 lines · Rene Herman
Quoted 6 lines · Andrew Morton
Quoted 4 lines · On Mon, 23 Apr 2007 14:32:36 +0200 Rene Herman <[email protected]> wrote:

Provide MODULE_MAINTAINER() as a convenient place to stick a name and email address both for drivers having multiple (current and non-current) authors and for when someone who wants to maintain a driver isn't so much an author.

[ snip ]

Quoted 2 lines

I'm not sure we want to do this - that's what ./MAINTAINERS is for and we end up having to maintain the same info in two places.

joe@user:~$ less ./MAINTAINERS
./MAINTAINERS: No such file or directory

MAINTAINERS is a developers thing, not users, yet a maintainer is someone who other than by developers wants to be contacted by users of a particular driver. Right now, a module exports a set of name and email addresses through the MODULE_AUTHOR tag but given multiple current and non-current authors, completely or largely orphaned drivers (I have a lot of junk PC hardware so I come across those relatively often) and people who might be interested in taking care of a driver but who do not consider themselves an author for (upto now) having done a s/, struct pt_regs// on it, that tag only confuses the issue of whom to contact.

And it in fact even does so when Joe does know about a MAINTAINERS file and does happen to have a kernel source tree lying around somewhere. With one set of addresses displayed prominently inside the sourcecode of the very driver and another one of in a MAINTAINERS file, the first one wins. Joe would have to be very new to Linux to trust something in the tree that's not actually compiled over something that is.

As the first response in this thread Cristoph Hellwig stated that MODULE_AUTHOR serves no purpose other than what MODULE_MAINTAINER would be serving. Others agreed and Adrian Bunk suggested deleting MODULE_AUTHOR outright.

That would actually also serve my purposes; if there's no MODULE_AUTHOR confusing the issue, I don't so much need a MODULE_MAINTAINER to fix it again. I believe having "modinfo" (optionally!) display a contact address for a driver might be a user advantage, but with all the wrong addresses gone, I don't really care deeply; MODULE_AUTHOR doesn't serve the purpose today and with it gone the user at least knows he needs to look elsewhere. MODULE_AUTHOR is also a credits issue but the information can be transferred to copyright headers. It would obviously also fix any possible maintenance issues.

Quoted 7 lines
Quoted 2 lines
Quoted 2 lines

Alan Cox believes that having author information embedded in the module serves a legal purpose though and objects to removal.

Wouldn't a /* comment */ satisfy AUTHOR needs?

It gives deserved attribution and should serve legal purpose just as well as a macro does (IANAL!).

Alan's opinion in [1] sounds reasonable (and I trust that he knows what he is talking about).

Quoted 3 lines

IMO we want MAINTAINER info in the macro and in modinfo, so I'm for removing MODULE_AUTHOR and just having MAINTAINER.
...

I don't think we want to expose maintainership information to users at all:
- duplicates information in MAINTAINERS
- maintainers sometimes disappear
- the 3 year old kernel of your distribution would contain 3 year old maintainership information

IMHO the default should be that users report problems with distribution kernels to their distribution and problems with ftp.kernel.org kernels to either linux-kernel or the kernel Bugzilla.

Quoted 1 line

~Randy

cu
Adrian

[1] http://lkml.org/lkml/2007/4/4/260

Signature · 4 lines
       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed
#32 lore raw
Quoted 56 lines · On Thu, 26 Apr 2007 17:52:06 +0200 Adrian Bunk wrote:
Quoted 55 lines · Randy Dunlap
Quoted 52 lines · On Thu, 26 Apr 2007 15:54:26 +0200 Adrian Bunk wrote:
Quoted 51 lines · Rene Herman
Quoted 6 lines · Andrew Morton
Quoted 4 lines · On Mon, 23 Apr 2007 14:32:36 +0200 Rene Herman <[email protected]> wrote:

Provide MODULE_MAINTAINER() as a convenient place to stick a name and email address both for drivers having multiple (current and non-current) authors and for when someone who wants to maintain a driver isn't so much an author.

[ snip ]

Quoted 2 lines

I'm not sure we want to do this - that's what ./MAINTAINERS is for and we end up having to maintain the same info in two places.

joe@user:~$ less ./MAINTAINERS
./MAINTAINERS: No such file or directory

MAINTAINERS is a developers thing, not users, yet a maintainer is someone who other than by developers wants to be contacted by users of a particular driver. Right now, a module exports a set of name and email addresses through the MODULE_AUTHOR tag but given multiple current and non-current authors, completely or largely orphaned drivers (I have a lot of junk PC hardware so I come across those relatively often) and people who might be interested in taking care of a driver but who do not consider themselves an author for (upto now) having done a s/, struct pt_regs// on it, that tag only confuses the issue of whom to contact.

And it in fact even does so when Joe does know about a MAINTAINERS file and does happen to have a kernel source tree lying around somewhere. With one set of addresses displayed prominently inside the sourcecode of the very driver and another one of in a MAINTAINERS file, the first one wins. Joe would have to be very new to Linux to trust something in the tree that's not actually compiled over something that is.

As the first response in this thread Cristoph Hellwig stated that MODULE_AUTHOR serves no purpose other than what MODULE_MAINTAINER would be serving. Others agreed and Adrian Bunk suggested deleting MODULE_AUTHOR outright.

That would actually also serve my purposes; if there's no MODULE_AUTHOR confusing the issue, I don't so much need a MODULE_MAINTAINER to fix it again. I believe having "modinfo" (optionally!) display a contact address for a driver might be a user advantage, but with all the wrong addresses gone, I don't really care deeply; MODULE_AUTHOR doesn't serve the purpose today and with it gone the user at least knows he needs to look elsewhere. MODULE_AUTHOR is also a credits issue but the information can be transferred to copyright headers. It would obviously also fix any possible maintenance issues.

Alan Cox believes that having author information embedded in the module serves a legal purpose though and objects to removal.

Quoted 7 lines
Quoted 4 lines

Wouldn't a /* comment */ satisfy AUTHOR needs?

It gives deserved attribution and should serve legal purpose just as well as a macro does (IANAL!).

Alan's opinion in [1] sounds reasonable (and I trust that he knows what he is talking about).

OK, I see.

Quoted 11 lines
Quoted 3 lines

IMO we want MAINTAINER info in the macro and in modinfo, so I'm for removing MODULE_AUTHOR and just having MAINTAINER.
...

I don't think we want to expose maintainership information to users at all:
- duplicates information in MAINTAINERS
- maintainers sometimes disappear
- the 3 year old kernel of your distribution would contain 3 year old maintainership information

Quoted 3 lines

IMHO the default should be that users report problems with distribution kernels to their distribution and problems with ftp.kernel.org kernels to either linux-kernel or the kernel Bugzilla.

s/linux-kernel/the appropriate mailing list/ please :)

so looks to me like we maintain the status quo.

Quoted 6 lines
Quoted 1 line

~Randy

cu
Adrian

[1] http://lkml.org/lkml/2007/4/4/260

---
~Randy
*** Remember to use Documentation/SubmitChecklist when testing your code ***

#33 lore raw
Quoted 7 lines · Randy Dunlap

...

Quoted 3 lines

IMHO the default should be that users report problems with distribution kernels to their distribution and problems with ftp.kernel.org kernels to either linux-kernel or the kernel Bugzilla.

s/linux-kernel/the appropriate mailing list/ please :)
...

Yes, in an ideal world.

But a user reporting an Oops might not even know where the problem lies.

And a user reporting a networking bug to linux-kernel hasn't really done any mistake. Forwarding bugs to the right people (as Andrew does and I'm sometimes doing) is actually an easy part of bug handling.

Quoted 1 line

~Randy

cu
Adrian

Signature · 4 lines
       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed
#34 lore raw
In reply to Adrian Bunk
Quoted 6 lines · Adrian Bunk

I don't think we want to expose maintainership information to users at all:
- duplicates information in MAINTAINERS
- maintainers sometimes disappear
- the 3 year old kernel of your distribution would contain 3 year old maintainership information

Right. Summary: adding maintainer info to source code and/or compiled modules doesn't seem to make any sense.

Quoted 3 lines

IMHO the default should be that users report problems with distribution kernels to their distribution and problems with ftp.kernel.org kernels to either linux-kernel or the kernel Bugzilla.

Yep. But it's really easy to miss something in lkml and/or in more specific lists, and the Bugzilla isn't email. That means we need a WWW-browseable database, with:
a) maintainers registrations with a simple email verification or something like that
b) possibility for registered people to declare they now maintain (or not) specific portions of the kernel
c) possibility for anyone to retrieve this information.

Other options?

Signature · 1 line
Krzysztof Halasa
#35 lore raw
Quoted 15 lines · Krzysztof Halasa

Adrian Bunk <[email protected]> writes:
...

Quoted 3 lines

IMHO the default should be that users report problems with distribution kernels to their distribution and problems with ftp.kernel.org kernels to either linux-kernel or the kernel Bugzilla.

Yep. But it's really easy to miss something in lkml and/or in more specific lists, and the Bugzilla isn't email. That means we need a WWW-browseable database, with:
a) maintainers registrations with a simple email verification or something like that
b) possibility for registered people to declare they now maintain (or not) specific portions of the kernel
c) possibility for anyone to retrieve this information.

Quoted 1 line

Other options?

What exactly is missing that the kernel Bugzilla doesn't already offer?

Quoted 1 line

Krzysztof Halasa

cu
Adrian

Signature · 4 lines
       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed
#36 lore raw
Quoted 1 line · Adrian Bunk

What exactly is missing that the kernel Bugzilla doesn't already offer?

Basically... unless I'm mistaken, nothing of the sort exists there.

Bugzilla is a database of bugs. What is needed is a database of people, mailing lists and some network resources.

How do I easily retrieve maintainer's email address for a given subsystem from Bugzilla?

Signature · 1 line
Krzysztof Halasa
#37 lore raw
Quoted 11 lines · Krzysztof Halasa
Quoted 1 line · Adrian Bunk

What exactly is missing that the kernel Bugzilla doesn't already offer?

Basically... unless I'm mistaken, nothing of the sort exists there.

Bugzilla is a database of bugs. What is needed is a database of people, mailing lists and some network resources.

How do I easily retrieve maintainer's email address for a given subsystem from Bugzilla?

Ah, now I get what you want.

You only get this information from MAINTAINERS plus knowledge who recently touched which code (git helps with the latter).

The problem with such a database would be the same as with the MAINTAINERS file: The information becomes outdated, and someone has to maintain it.

Sending a patch against MAINTAINERS is easy - I don't see a WWW-browseable database being in any respect better than the MAINTAINERS file.

Quoted 1 line

Krzysztof Halasa

cu
Adrian

Signature · 4 lines
       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed
#38 lore raw
Quoted 3 lines · Adrian Bunk

The problem with such a database would be the same as with the MAINTAINERS file: The information becomes outdated, and someone has to maintain it.

Sure, I can easily grep .../linux-current/MAINTAINERS. But I think the whole problem is with people who don't have current sources.

Signature · 1 line
Krzysztof Halasa
#39 lore raw
Quoted 8 lines · Krzysztof Halasa
Quoted 3 lines · Adrian Bunk

The problem with such a database would be the same as with the MAINTAINERS file: The information becomes outdated, and someone has to maintain it.

Sure, I can easily grep .../linux-current/MAINTAINERS. But I think the whole problem is with people who don't have current sources.

No, even MAINTAINERS in the latest sources always contains outdated entries and lacks information.

If you think "no current sources" is a problem that should be solved, the solution would simply be a link to [1].

Quoted 1 line

Krzysztof Halasa

cu
Adrian

[1] http://git.kernel.org/gitweb.cgi?p=linux/kernel/git/torvalds/linux-2.6.git;a=blob;f=MAINTAINERS;hb=HEAD

Signature · 4 lines
       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed
#40 lore raw
Quoted 2 lines · Adrian Bunk

No, even MAINTAINERS in the latest sources always contains outdated entries and lacks information.

Sure, but that can't be corrected using technical meanings.

Quoted 4 lines

If you think "no current sources" is a problem that should be solved, the solution would simply be a link to [1].

[1] http://git.kernel.org/gitweb.cgi?p=linux/kernel/git/torvalds/linux-2.6.git;a=blob;f=MAINTAINERS;hb=HEAD

Well, the most simple solutions are sometimes quite good.

Signature · 1 line
Krzysztof Halasa
#41 lore raw
In reply to Adrian Bunk
Quoted 4 lines · Adrian Bunk

The problem with such a database would be the same as with the MAINTAINERS file: The information becomes outdated, and someone has to maintain it.

Sending a patch against MAINTAINERS is easy - I don't see a WWW-browseable database being in any respect better than the MAINTAINERS file.

Well, in fact, the database could _be_ (a parsed version of) the current MAINTAINERS file; it's already in a strict, parseable format. I know that MAINTAINERS itself is not very good at staying up to date, but making more intensive use of it like that might in fact improve that...

Being able to have "modinfo module" then point at a specific link would require a bit more though. It could be agreed upon to, say, have a module that the kernel installs as

	/lib/modules/$(uname -r)/kernel/drivers/cdrom/mitsumi.ko

have a maintainer page at

http://www.kernel.org/maintainer/driver/cdrom/mitsumi.html

but to generate that page, the maintainer file would then need to grow a way to associate the "MITSUMI CDROM DRIVER" entry (don't bother, it's not there) in the MAINTAINERS file with that page (and simply listing that install path in the entry isn't good -- I believe you would wan't to pin down a structure like that).

Anyways -- might all also be a bit of overkill anyway. If I can get those email adresses from MODULE_AUTHOR out of the way, I'm mostly happy.

Rene.

#43 lore raw
Quoted 4 lines · Rene Herman
Quoted 1 line · Adrian Bunk

What exactly is missing that the kernel Bugzilla doesn't already offer?

Users?

That is the best answer of all, and I've stated my objections to that very nearly worthless utility before. And that is exactly why it doesn't have "Users". I have never, ever gone to bugzilla without killing well over an hour convincing it that yes, I really want DO to enter a bug report, not search for an old one. Let me simply say that such and such doesn't work, attach the dmesg (or whatever) snip to show how it failed, and get on with it. Wasting 2 hours crawling around looking for loopholes in the fence so I can wrap the report around a rock and throw it over? That, and 30" of rain will raise 180 bushels to the acre where I come from...

Signature · 5 lines
Cheers, Gene
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Statistics means never having to say you're certain.
#44 lore raw
Quoted 6 lines · Gene Heskett
Quoted 4 lines · Rene Herman
Quoted 1 line · Adrian Bunk

What exactly is missing that the kernel Bugzilla doesn't already offer?

Users?

Quoted 7 lines

That is the best answer of all, and I've stated my objections to that very nearly worthless utility before. And that is exactly why it doesn't have "Users". I have never, ever gone to bugzilla without killing well over an hour convincing it that yes, I really want DO to enter a bug report, not search for an old one. Let me simply say that such and such doesn't work, attach the dmesg (or whatever) snip to show how it failed, and get on with it.

[...]

Of course it is polite (and more or less expected from reporters, independently of whether bugzilla or a mailinglist or personal e-mail is used for a report) to search for similar reports and join them. However, it doesn't stop there --- a bugzilla entry into which a lot of different bugs are entered (because the reporters mistook them for the same or didn't realize they face a number of separate problems) becomes hard to handle for maintainers/ developers.

So, it is actually easier after all to have users create *duplicate* bugs and then simply mark them as such once identified. Bugzilla directly supports tracking of duplicates while my mail user agent doesn't. (Some advanced MUAs might do, by tagging or the like. But that would only be visible to myself.)

As subsystem maintainer, I am actively using kernel.org's bugzilla and I am polling more or less (mostly: /less/) frequently some distributor bugzillas. Here are my experiences:

- I can't work without kernel.org's bugzilla simply because the average bug fixing rate of "my" subsystem is so slow. I personally need a tracker better than some yellow sticky notes or whatever, and I chose kernel.org's bugzilla for it.

That's one reason why I once and again ask people who report bugs at the -devel or -user mailinglist to enter their bug into bugzilla. But often I request this only after an initial conversation with the reporter, i.e. after the issue was already clarified a bit. Side effects are that the effort of the user to enter the report becomes minimal (or even zero if I do it myself instead), and that the new entry is already rather qualified.

Actually, a good deal of "my" currently unresolved bugs at bugzilla.kernel.org were entered by myself. Some of those bugs are behind-the-scene bugs though which only indirectly affect endusers, thus wouldn't have reported by them anyway.

- Using distributor's bugzillas is a mixed bag. By nature, it takes more effort for kernel maintainers because linux-kernel is only one among a huge number of programs tracked in these bugzillas.

There were times were I was able to get a lot of highly useful reports out of one particular bugzilla; and I name it here because its admins and users did such a great job IME: Redhat's bugzilla. Many of the bug entries there were highly focused and qualified and up-to-date, and some helped to get to bugfixes soon. This positive experience somewhat diminished lately after the more prominent bugs were fixed.

At other distributor bugzillas, the usually encountered difficulties besides the broad scope of their trackers were, in no particular order:
- Lack of detail in bug descriptions,
- many loosely related or unrelated bugs under the same ticket,
- difficulties to work with the reporters to get more diagnostics (for varying reasons),
- long delay between upstream release of regressions and report of regressions by distribution users,
- my lacking polling frequency, being caused by and at the same time amplifying these other difficulties.

So, bugzilla.kernel.org and to a certain degree distributors' bugzillas are valuable to me. Still, switching between bugzilla and mailinglist sometimes becomes awkward, IME in a similar way like switching between mailinglist and personal e-mail. Also, "my" subsystem (ieee1394) almost doesn't have to deal anymore with new development, neither on the kernel side nor as far as available hardware is concerned. Therefore my findings certainly do not reflect what other subsystem maintainers are facing.

Back to MODULE_MAINTAINER and what Adrian said about where to report bugs:

Quoted 1 line

From the maintainer's point of view, my personal preference for bug

reports about "my" subsystem is, in descending order:
- The -devel or -user mailinglist (but often in combination with bugzilla),
- bugzilla,
- ...?,
- personal e-mail. (Just don't do it.)
But I do understand why the opener of this thread preferred personal e-mail; he has explained it multiple times now. He has valid points, although they don't apply to by far most of the kernel subsystems.

Signature · 3 lines
Stefan Richter
-=====-=-=== -=-- ==-==
http://arcgraph.de/sr/
#45 lore raw
In reply to Adrian Bunk
Quoted 1 line · Adrian Bunk

I don't think we want to expose maintainership information to users at all:

With the point you make about old installed kernel modules having outdated information forever you've in fact convinced me that MODULE_MAINTAINER is not a good idea. If I decide after some time that I no longer want to be a maintainer anymore, I can delete the MODULE_MAINTAINER and/or the entry in MAINTAINERS in the current kernel, but I could be getting mail for a long time still through old installs.

The problem I still have is that we _do_ expose authorship information, generally complete with (an) email address(es), which is the only address available from the binary and which is displayed on every invocation of the user tool "modinfo". Proper contact information may rather be available from a text file off in a source tree somewhere instead, but this information is ofcourse going to remain being interpreted as such however one may feel about it.

Deleting authorship information can't happen for legal reasons. Not in the global "delete them all" sense but also not in a local sense; a maintainer doesn't have the right to delete a MODULE_AUTHOR tag but not being able to additionally supply contact information _alongside_ that information then leaves the confusing message the tag sends as is.

Deleting the email addresses from the MODULE_AUTHOR tag would go some ways to fix it; it's then at least clearer that the author is not being displayed as a general contact for the driver. It may on the other hand want to remain as a legal contact and I only know of "modinfo" as a normal way of listing the tags, so even a very minimal solution such as having modinfo supress the author tag, or even just any email address in it, would be good enough.

Rene.

#46 lore raw
Quoted 1 line · Rene Herman

With the point you make about old installed kernel modules having outdated information forever you've in fact convinced me that MODULE_MAINTAINER is not a good idea.

[ ... ]

Quoted 2 lines

Deleting the email addresses from the MODULE_AUTHOR tag would go some ways to fix it; it's then at least clearer that the author is not being displayed as a general contact for the driver. It may on the other hand want to remain as a legal contact and I only know of "modinfo" as a normal way of listing the tags, so even a very minimal solution such as having modinfo supress the author tag, or even just any email address in it, would be good enough.

Ie, something like the attached minimal patch to modinfo that just supresses the author= tag from the default output; the information is still available from modinfo -a.

However. Looking at the MODULE_AUTHOR tags in the tree:

rene@7ixe4:~/src/linux/local$ grep -r MODULE_AUTHOR * | wc -l
2211

more than half of them already don't provide an email address:

rene@7ixe4:~/src/linux/local$ grep -r "MODULE_AUTHOR(.*<.*>.*)" * | wc -l
1088

Alan is one of the people using just MODULE_AUTHOR("Alan Cox") without an address. Given that the email address is all that I want to supress; how about just deleting that instead?

I'd prefer that; unlike the name, the adress is information that can get outdated and moreover, removing the address not only from the modinfo output but from the source directly means it can't be mistaken for a contact address there either.

Comments? Objections? If none, I'll start submitting patches removing email addresses from the MODULE_AUTHOR tags in the tree.

Rene.

modinfo.c renamed from modinfo.c.orig +3 -0
@@ -167,6 +167,9 @@
 	for (; info; info = next_string(info, &size)) {
 		char *eq, *colon;
 
+		if (strstarts(info, "author="))
+			continue;
+
 		/* We expect this in parm and parmtype. */
 		colon = strchr(info, ':');
 
#47 lore raw

Rene Herman <[email protected]> writes:

[MODULE_AUTHOR]

Quoted 2 lines

Given that the email address is all that I want to supress; how about just deleting that instead?

Makes sense at least WRT the "problematic" modules.

include/linux/module.h says:

/* Author, ideally of form NAME <EMAIL>[, NAME <EMAIL>]*[ and NAME <EMAIL>] */ #define MODULE_AUTHOR(_author) MODULE_INFO(author, _author)

I think we should get rid of the "EMAIL" comments:

/* Author, ideally of form FULL NAME [, FULL NAME ]*[ and FULL NAME] */ #define MODULE_AUTHOR(_author) MODULE_INFO(author, _author)

I'd rather let the respective maintainers/authors take care of "their" MODULE_AUTHOR entries (though the removal seems sane for me personally).

Quoted 2 lines

I'd prefer that; unlike the name, the adress is information that can get outdated

Right

Quoted 3 lines

and moreover, removing the address not only from the modinfo output but from the source directly means it can't be mistaken for a contact address there either.

Sure.

Signature · 1 line
Krzysztof Halasa
#48 lore raw
In reply to Robert P. J. Day
Quoted 9 lines · Robert P. J. Day
Quoted 7 lines · Rusty Russell
Quoted 3 lines · Rene Herman

On 04/04/2007 06:38 PM, Rene Herman wrote:

Rusty?

Valid points have been made on both sides. I suggest:

Quoted 8 lines
Quoted 2 lines

#define MODULE_MAINTAINER(_maintainer) \
MODULE_AUTHOR("(Maintained by) "_maintainer)

why bring MODULE_AUTHOR into it? just define it in terms of MODULE_INFO:

#define MODULE_MAINTAINER(_m) MODULE_INFO(_m, "(Maintained by)" \
maintainer)

whoops, the above is obviously syntactically incorrect, but you get the idea regarding MODULE_INFO, right?

rday

Signature · 7 lines
========================================================================
Robert P. J. Day
Linux Consulting, Training and Annoying Kernel Pedantry
Waterloo, Ontario, CANADA

http://fsdev.net/wiki/index.php?title=Main_Page
========================================================================
#49 lore raw
In reply to Robert P. J. Day
Quoted 7 lines · Robert P. J. Day
Quoted 5 lines · Rusty Russell
Quoted 3 lines · Rene Herman

On 04/04/2007 06:38 PM, Rene Herman wrote:

Rusty?

Quoted 7 lines
Quoted 4 lines

Valid points have been made on both sides. I suggest:

#define MODULE_MAINTAINER(_maintainer) \
MODULE_AUTHOR("(Maintained by) "_maintainer)

why bring MODULE_AUTHOR into it? just define it in terms of MODULE_INFO:

Because author is an established field. People might well search for it. This is fairly clear, and assuming that the maintainer has actually done any maintenance, they're an author too.

Rusty.