Hi all,
With the v3.0 name change I was looking over what might break, and I'm
seriously considering turning off the incremental diff robot on
kernel.org. It's not clear to me that it is actually useful anymore,
with git and all.
Do anyone actually use these anymore?
-hpa
--
H. Peter Anvin, Intel Open Source Technology Center
I work for Intel. I don't speak on their behalf.
On Mon, May 30, 2011 at 12:03:03PM -0700, H. Peter Anvin wrote:
> Hi all,
>
> With the v3.0 name change I was looking over what might break, and I'm
> seriously considering turning off the incremental diff robot on
> kernel.org. It's not clear to me that it is actually useful anymore,
> with git and all.
What is the incremental diff robot?
Does it make the patches between 2.6.x to 2.6.x+1?
>
> Do anyone actually use these anymore?
If the above is the case, yes I do. And we need to update the ketchup
script as well.
-- Steve
On Mon, 30 May 2011 15:45:00 -0400 Steven Rostedt wrote:
> On Mon, May 30, 2011 at 12:03:03PM -0700, H. Peter Anvin wrote:
> > Hi all,
> >
> > With the v3.0 name change I was looking over what might break, and I'm
> > seriously considering turning off the incremental diff robot on
> > kernel.org. It's not clear to me that it is actually useful anymore,
> > with git and all.
>
> What is the incremental diff robot?
>
> Does it make the patches between 2.6.x to 2.6.x+1?
>
> >
> > Do anyone actually use these anymore?
>
> If the above is the case, yes I do. And we need to update the ketchup
> script as well.
and scripts/patch-kernel needs to be fixed (or dropped).
It will use the incremental diffs, but I doubt that patch-kernel or the
incremental diffs model has many users (he said with no proof :).
They could go away IMO. I only use git or -rc or daily tarballs.
The only time that I have used incremental diffs was for a poor man's
bisect, but git does it much better.
---
~Randy
*** Remember to use Documentation/SubmitChecklist when testing your code ***
On 05/30/2011 12:45 PM, Steven Rostedt wrote:
> On Mon, May 30, 2011 at 12:03:03PM -0700, H. Peter Anvin wrote:
>> Hi all,
>>
>> With the v3.0 name change I was looking over what might break, and I'm
>> seriously considering turning off the incremental diff robot on
>> kernel.org. It's not clear to me that it is actually useful anymore,
>> with git and all.
>
> What is the incremental diff robot?
>
> Does it make the patches between 2.6.x to 2.6.x+1?
>
No, but it generates the patches between, say, 2.6.x.y and 2.6.x.y+1 or
2.6.x-rcy and 2.6.x-rcy+1. Anything in an "incr" directory.
-hpa
--
H. Peter Anvin, Intel Open Source Technology Center
I work for Intel. I don't speak on their behalf.
On Mon, May 30, 2011 at 01:49:01PM -0700, H. Peter Anvin wrote:
> On 05/30/2011 12:45 PM, Steven Rostedt wrote:
> > On Mon, May 30, 2011 at 12:03:03PM -0700, H. Peter Anvin wrote:
> >> Hi all,
> >>
> >> With the v3.0 name change I was looking over what might break, and I'm
> >> seriously considering turning off the incremental diff robot on
> >> kernel.org. It's not clear to me that it is actually useful anymore,
> >> with git and all.
> >
> > What is the incremental diff robot?
> >
> > Does it make the patches between 2.6.x to 2.6.x+1?
> >
>
> No, but it generates the patches between, say, 2.6.x.y and 2.6.x.y+1 or
> 2.6.x-rcy and 2.6.x-rcy+1. Anything in an "incr" directory.
Yep, I use those incremental patches over here.
On Tue, 2011-05-31 at 14:38 +0200, Heiko Carstens wrote:
> > No, but it generates the patches between, say, 2.6.x.y and 2.6.x.y+1 or
> > 2.6.x-rcy and 2.6.x-rcy+1. Anything in an "incr" directory.
>
> Yep, I use those incremental patches over here.
And the ketchup utility does too. But that needs to be fixed as well.
Luckily I just handed over maintainership of ketchup to someone else :)
-- Steve
Hi!
> With the v3.0 name change I was looking over what might break, and I'm
> seriously considering turning off the incremental diff robot on
> kernel.org. It's not clear to me that it is actually useful anymore,
> with git and all.
>
> Do anyone actually use these anymore?
Yes.
Usecase one is zaurus, which is just not powerful enough for git, and
usecase two are companies that only work with releases... ketchup is
very useful for both.
Pavel
--
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
Hi Peter,
On Mon, May 30, 2011 at 12:03:03PM -0700, H. Peter Anvin wrote:
> Hi all,
>
> With the v3.0 name change I was looking over what might break, and I'm
> seriously considering turning off the incremental diff robot on
> kernel.org. It's not clear to me that it is actually useful anymore,
> with git and all.
>
> Do anyone actually use these anymore?
Well, I don't think the incremental patches are *that much* used.
Speaking for my case, my longterm release scripts do produce them
anyway since it's trivial with git, and I suppose many others do
as well.
I suggest that we disable the robot and wait for any possible
complaints which are to be expected anyway with a version change.
Then it'll be more productive to help tree owners to adapt their
scripts than it would be to keep the robot running.
Just my 2 cents,
Willy
On Tuesday 31 May 2011 21:43:59 Pavel Machek wrote:
> > With the v3.0 name change I was looking over what might break, and I'm
> > seriously considering turning off the incremental diff robot on
> > kernel.org. It's not clear to me that it is actually useful anymore,
> > with git and all.
Are you talking about just ftp://ftp.kernel.org/pub/linux/kernel/*/incr/
or also about ftp://ftp.kernel.org/pub/linux/kernel/v2.6/patch-* and
Changelog-*?
I can see reasons to keep the latter, but not necessarily the former.
> > Do anyone actually use these anymore?
>
> Yes.
>
> Usecase one is zaurus, which is just not powerful enough for git, and
> usecase two are companies that only work with releases... ketchup is
> very useful for both.
For usecase one, is it possible to replace the download links with
links into http://git.kernel.org/ ?
For usecase two, why wouldn't those be able to just download the entire
release tarball? At least, I would expect that there is no need for
the small incremental patches in their case, only for the full
inter-release patches.
Arnd
On 05/31/2011 12:52 PM, Willy Tarreau wrote:
>
> Well, I don't think the incremental patches are *that much* used.
> Speaking for my case, my longterm release scripts do produce them
> anyway since it's trivial with git, and I suppose many others do
> as well.
>
> I suggest that we disable the robot and wait for any possible
> complaints which are to be expected anyway with a version change.
> Then it'll be more productive to help tree owners to adapt their
> scripts than it would be to keep the robot running.
>
Yes, this would be one option. It would at least save the effort of
chasing problems when it breaks due to tree-owner changes.
-hpa
On 05/31/2011 01:04 PM, Arnd Bergmann wrote:
>
> Are you talking about just ftp://ftp.kernel.org/pub/linux/kernel/*/incr/
> or also about ftp://ftp.kernel.org/pub/linux/kernel/v2.6/patch-* and
> Changelog-*?
>
> I can see reasons to keep the latter, but not necessarily the former.
>
The former only.
>>
>> Usecase one is zaurus, which is just not powerful enough for git, and
>> usecase two are companies that only work with releases... ketchup is
>> very useful for both.
>
> For usecase one, is it possible to replace the download links with
> links into http://git.kernel.org/ ?
>
> For usecase two, why wouldn't those be able to just download the entire
> release tarball? At least, I would expect that there is no need for
> the small incremental patches in their case, only for the full
> inter-release patches.
>
Zaurus I don't consider too important; anyone with a Zaurus is
realistically going to have a real development platform on hand.
-hpa
On Tue 2011-05-31 13:13:51, H. Peter Anvin wrote:
> On 05/31/2011 01:04 PM, Arnd Bergmann wrote:
> >
> > Are you talking about just ftp://ftp.kernel.org/pub/linux/kernel/*/incr/
> > or also about ftp://ftp.kernel.org/pub/linux/kernel/v2.6/patch-* and
> > Changelog-*?
> >
> > I can see reasons to keep the latter, but not necessarily the former.
>
> The former only.
Aha, I guess I have to agree... /incr/ is not terribly
important. Anyone doing _that_ kind of development probably has git.
Pavel
--
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
On Mon 2011-05-30 13:49:01, H. Peter Anvin wrote:
> On 05/30/2011 12:45 PM, Steven Rostedt wrote:
> > On Mon, May 30, 2011 at 12:03:03PM -0700, H. Peter Anvin wrote:
> >> Hi all,
> >>
> >> With the v3.0 name change I was looking over what might break, and I'm
> >> seriously considering turning off the incremental diff robot on
> >> kernel.org. It's not clear to me that it is actually useful anymore,
> >> with git and all.
> >
> > What is the incremental diff robot?
> >
> > Does it make the patches between 2.6.x to 2.6.x+1?
> >
>
> No, but it generates the patches between, say, 2.6.x.y and 2.6.x.y+1 or
> 2.6.x-rcy and 2.6.x-rcy+1. Anything in an "incr" directory.
-git1 to -git2 diffs are probably not useful.
2.6.39.1 to 2.6.39.2 diffs are useful IMO - people using stable
kernels do not use git.
I'm using -rc1 to -rc2 diffs. Useful :-). Pavel
--
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
On Mon, May 30, 2011 at 12:03:03PM -0700, H. Peter Anvin wrote:
> Hi all,
>
> With the v3.0 name change I was looking over what might break, and I'm
> seriously considering turning off the incremental diff robot on
> kernel.org. It's not clear to me that it is actually useful anymore,
> with git and all.
>
> Do anyone actually use these anymore?
I use them to go from one stable version to the next and it is quite
convenient. Maybe there is another way to achieve that (that "all" in "git and
all")?
Thanks,
Tvrtko
Hi Peter,
On Mon, 30 May 2011 12:03:03 -0700, H. Peter Anvin wrote:
> With the v3.0 name change I was looking over what might break, and I'm
> seriously considering turning off the incremental diff robot on
> kernel.org. It's not clear to me that it is actually useful anymore,
> with git and all.
>
> Do anyone actually use these anymore?
Do you mean files in testing/incr such as patch-2.6.39-rc6-rc7.bz2?
Yes, I am still using these. That being said:
* I guess I could extract these myself from git with a simple git
diff command?
* I don't give a damn to older incremental patches, only to the ones of
the current development cycle.
So if you want to delete all the old files, I'm all for it, and if you
decide to no longer generate the incremental patches at all, I can live
with it.
--
Jean Delvare
On Wed, Jun 1, 2011 at 4:09 PM, Jean Delvare <[email protected]> wrote:
> Do you mean files in testing/incr such as patch-2.6.39-rc6-rc7.bz2?
> Yes, I am still using these. That being said:
> * I guess I could extract these myself from git with a simple git
> ?diff command?
git diff v2.6.39-rc6..v2.6.39-rc7
On 06/01/2011 01:43 AM, Tvrtko Ursulin wrote:
>
> I use them to go from one stable version to the next and it is quite
> convenient. Maybe there is another way to achieve that (that "all" in "git and
> all")?
>
You can pull down the stable git tree and do diffs to your heart's
content, obviously.
-hpa
--
H. Peter Anvin, Intel Open Source Technology Center
I work for Intel. I don't speak on their behalf.