| Message ID | 1352464732-22984-1-git-send-email-alexander@mezon.ru |
|---|---|
| State | Accepted |
| Commit | 64e04fb27a9a0a4fcc519bcc7344a67009490a91 |
| Headers | show |
Alexander, On Fri, 9 Nov 2012 16:38:52 +0400, Alexander Khryukin wrote: > -LIBTOOL_VERSION = 2.2.10 > +LIBTOOL_VERSION = 2.4.2 How much testing did you give to this version bump? A libtool version bump is a very sensitive operation, as host-libtool is used to autoreconfigure many packages in Buildroot. Therefore, this patch needs a good amount of testing before being committed (ideally testing that all packages having <foo>_AUTORECONF = YES still build). Thanks, Thomas
2012/11/10 Thomas Petazzoni <thomas.petazzoni@free-electrons.com> > Alexander, > > On Fri, 9 Nov 2012 16:38:52 +0400, Alexander Khryukin wrote: > > > -LIBTOOL_VERSION = 2.2.10 > > +LIBTOOL_VERSION = 2.4.2 > > How much testing did you give to this version bump? A libtool version > bump is a very sensitive operation, as host-libtool is used to > autoreconfigure many packages in Buildroot. Therefore, this patch needs > a good amount of testing before being committed (ideally testing that > all packages having <foo>_AUTORECONF = YES still build). > > Thanks, > > Thomas > -- > Thomas Petazzoni, Free Electrons > Kernel, drivers, real-time and embedded Linux > development, consulting, training and support. > http://free-electrons.com > With new libtool i built my system many times and it works.
On 11/10/12 10:53, Thomas Petazzoni wrote: > Alexander, > > On Fri, 9 Nov 2012 16:38:52 +0400, Alexander Khryukin wrote: > >> -LIBTOOL_VERSION = 2.2.10 >> +LIBTOOL_VERSION = 2.4.2 > > How much testing did you give to this version bump? A libtool version > bump is a very sensitive operation, as host-libtool is used to > autoreconfigure many packages in Buildroot. Therefore, this patch needs > a good amount of testing before being committed (ideally testing that > all packages having<foo>_AUTORECONF = YES still build). Can't we rely on the autobuilders to do that? That said, I wouldn't do this for 2012.11 anymore... Regards, Arnout
On Sun, 11 Nov 2012 23:00:56 +0100, Arnout Vandecappelle wrote: > On 11/10/12 10:53, Thomas Petazzoni wrote: > > Alexander, > > > > On Fri, 9 Nov 2012 16:38:52 +0400, Alexander Khryukin wrote: > > > >> -LIBTOOL_VERSION = 2.2.10 > >> +LIBTOOL_VERSION = 2.4.2 > > > > How much testing did you give to this version bump? A libtool > > version bump is a very sensitive operation, as host-libtool is used > > to autoreconfigure many packages in Buildroot. Therefore, this > > patch needs a good amount of testing before being committed > > (ideally testing that all packages having<foo>_AUTORECONF = YES > > still build). > > Can't we rely on the autobuilders to do that? Sure, we'll certainly rely on the autobuilders for a complete testing. But I wanted to know if it had been tested again 2 packages or 20-40 packages, which makes quite a bit of difference :) > That said, I wouldn't do this for 2012.11 anymore... For sure, it should not be part of 2012.11, we already have enough issues to fix. Thomas
Am 11.11.2012 23:05, schrieb Thomas Petazzoni: > On Sun, 11 Nov 2012 23:00:56 +0100, Arnout Vandecappelle wrote: >> On 11/10/12 10:53, Thomas Petazzoni wrote: >>> Alexander, >>> >>> On Fri, 9 Nov 2012 16:38:52 +0400, Alexander Khryukin wrote: >>> >>>> -LIBTOOL_VERSION = 2.2.10 >>>> +LIBTOOL_VERSION = 2.4.2 >>> How much testing did you give to this version bump? A libtool >>> version bump is a very sensitive operation, as host-libtool is used >>> to autoreconfigure many packages in Buildroot. Therefore, this >>> patch needs a good amount of testing before being committed >>> (ideally testing that all packages having<foo>_AUTORECONF = YES >>> still build). >> Can't we rely on the autobuilders to do that? > Sure, we'll certainly rely on the autobuilders for a complete testing. Hello, did you ever think about inventing a kind of "testing" branch with a seperate autobuilder running on? That could keep major build problems away from the master branch. Just my 2 cents Stephan > But I wanted to know if it had been tested again 2 packages or 20-40 > packages, which makes quite a bit of difference :) > >> That said, I wouldn't do this for 2012.11 anymore... > For sure, it should not be part of 2012.11, we already have enough > issues to fix. > > Thomas
2012/11/12 Stephan Hoffmann <sho@relinux.de> > Am 11.11.2012 23:05, schrieb Thomas Petazzoni: > > On Sun, 11 Nov 2012 23:00:56 +0100, Arnout Vandecappelle wrote: > >> On 11/10/12 10:53, Thomas Petazzoni wrote: > >>> Alexander, > >>> > >>> On Fri, 9 Nov 2012 16:38:52 +0400, Alexander Khryukin wrote: > >>> > >>>> -LIBTOOL_VERSION = 2.2.10 > >>>> +LIBTOOL_VERSION = 2.4.2 > >>> How much testing did you give to this version bump? A libtool > >>> version bump is a very sensitive operation, as host-libtool is used > >>> to autoreconfigure many packages in Buildroot. Therefore, this > >>> patch needs a good amount of testing before being committed > >>> (ideally testing that all packages having<foo>_AUTORECONF = YES > >>> still build). > >> Can't we rely on the autobuilders to do that? > > Sure, we'll certainly rely on the autobuilders for a complete testing. > Hello, > > did you ever think about inventing a kind of "testing" branch with a > seperate autobuilder running on? That could keep major build problems > away from the master branch. > > Just my 2 cents > > Stephan > > But I wanted to know if it had been tested again 2 packages or 20-40 > > packages, which makes quite a bit of difference :) > > > >> That said, I wouldn't do this for 2012.11 anymore... > > For sure, it should not be part of 2012.11, we already have enough > > issues to fix. > > > > Thomas > > > -- > reLinux - Stephan Hoffmann > Am Schmidtgrund 124 50765 Köln > Tel. +49.221.95595-19 Fax: -64 > www.reLinux.de sho@reLinux.de > > > _______________________________________________ > buildroot mailing list > buildroot@busybox.net > http://lists.busybox.net/mailman/listinfo/buildroot > I tested new libtool for all packages in Develop section + a lot of others. Also i can share own .config
Stephan, On Mon, 12 Nov 2012 08:46:26 +0100, Stephan Hoffmann wrote: > did you ever think about inventing a kind of "testing" branch with a > seperate autobuilder running on? That could keep major build problems > away from the master branch. Yes, this has been discussed during the Developers Meeting, at least for the "next" branch that gets created once -rc1 is out, to accumulate patches for the next release. Thomas
On 11/12/12 11:26, Thomas Petazzoni wrote: > Stephan, > > On Mon, 12 Nov 2012 08:46:26 +0100, Stephan Hoffmann wrote: > >> did you ever think about inventing a kind of "testing" branch with a >> seperate autobuilder running on? That could keep major build problems >> away from the master branch. > > Yes, this has been discussed during the Developers Meeting, at least > for the "next" branch that gets created once -rc1 is out, to accumulate > patches for the next release. Actually it does make sense to have it in general. We could allow more committers on that branch, and Peter could cherry-pick from it, which could save him a few seconds per patch. And it could make master slightly more stable, because only patches that don't kill the autobuilders would get cherry-picked. Maybe something to discuss in February. Regards, Arnout
diff --git a/package/libtool/libtool.mk b/package/libtool/libtool.mk index 4b9657c..ae7a4b8 100644 --- a/package/libtool/libtool.mk +++ b/package/libtool/libtool.mk @@ -3,7 +3,7 @@ # libtool # ############################################################# -LIBTOOL_VERSION = 2.2.10 +LIBTOOL_VERSION = 2.4.2 LIBTOOL_SOURCE = libtool-$(LIBTOOL_VERSION).tar.gz LIBTOOL_SITE = $(BR2_GNU_MIRROR)/libtool LIBTOOL_INSTALL_STAGING = YES
Signed-off-by: Alexander Khryukin <alexander@mezon.ru> --- package/libtool/libtool.mk | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)