Re: major fixes to backport.

[Loïc Minier <lool+gnome via ecp fr>]

        Hi,

 Thanks a lot for compiling this list!  Here are some remarks to 2.14
 applicability (Debian will soon ship a new stable release with G2.14).

On Sat, Feb 24, 2007, Andre Klapper wrote:
> ==========
> 2006-10-30, nautilus: 1704 rejected traces on 2007-01-23, 
> 2716 rejected traces on 2007-02-24. so this is still a HUGE problem.
> http://bugzilla.gnome.org/show_bug.cgi?id=352592#c258
> http://svn.gnome.org/viewcvs/nautilus/trunk/src/file-manager/fm-icon-view.c?r1=12557&r2=12556&pathrev=12557

 SVN.g.o was down so I didn't really check the actual commit but I
 attached a backport to 2.14 of the 2.16.2 -> .3 diff.

 The diff also shows an interesting double free() bug which I've also
 pulled, see #321175.

> ==========
> 2006-11-01, nautilus: 754 rejected traces on 2007-01-23,
> 1164 rejected traces on 2007-02-24.
> http://bugzilla.gnome.org/show_bug.cgi?id=356672#c113

 Patch applies under 2.14.

> ==========
> 2006-11-02, nautilus: 474 rejected traces on 2007-01-23,
> 694 rejected traces on 2007-02-24.
> http://bugzilla.gnome.org/show_bug.cgi?id=348161#c101

 Unfortunately, too intrusive changes in the 2.16 cycle to backport this
 to 2.14.

> ==========
> 2006-12-13, at-spi: 424 dups
> http://bugzilla.gnome.org/show_bug.cgi?id=377079#c72

 Not backportable to 2.14 as it requires Gtk 2.10.

> ==========
> 2007-01-06, control-center: 189 dups
> http://bugzilla.gnome.org/show_bug.cgi?id=356435#c46 #c53 #c54

 Applies to 2.14; backport attached.

> ==========
> 2007-01-11, nautilus: 236 dups
> http://bugzilla.gnome.org/show_bug.cgi?id=320020#c28

 Attached the commit and the 2.14 backport.

> ==========
> 2007-01-17, epiphany [sic!]: 113 dups
> http://bugzilla.gnome.org/show_bug.cgi?id=351814#c93

 Attached the commit and the 2.14 backport.

> ==========
> 2007-01-22, gnome-terminal: 251 dups
> http://bugzilla.gnome.org/show_bug.cgi?id=353498#c191
>   http://bugzilla.gnome.org/show_bug.cgi?id=399282#c3

 Attached the GNOME 2.14 backport (vte 0.12.x).

> ==========
> 2007-01-24, gst-plugins-ugly: 349 dups
> http://bugzilla.gnome.org/show_bug.cgi?id=336370#c67

 I did not attach any patch since this isn't truly GNOME, but this
 applied against 0.10.4.

> ==========
> 2007-02-11, gnome-system-tools: 359 dups
> http://bugzilla.gnome.org/show_bug.cgi?id=356180#c136
>   http://bugzilla.gnome.org/show_bug.cgi?id=354536
>   http://svn.gnome.org/viewcvs/gnome-system-tools?view=rev&revision=3765

 Does not seem to affect 2.14.

> ==========
> 2007-02-24, gedit: 164 dups
> http://bugzilla.gnome.org/show_bug.cgi?id=354046#c184
>   http://svn.gnome.org/viewcvs/gedit?view=rev&revision=5511

 Attached commit and GNOME 2.14 backport.

> i kindly ask backport-policy vendors/distros to add a short comment to
> the bugzilla.gnome.org report if/when a patch has been backported
> (HIGHLY appreciated), so it becomes easier to track whether a patch has
> fixed an issue or not. would be also great if this could be done for
> this email here, but i know that my list is looong...
> we cannot always "wait for the next major release", this is way too much
> workload for the bugsquad, and you also will not get content users by
> just sitting out the instability problems...

 Will do my best; not sure of the actual form it will take though.

 FYI, all of the above uploaded to Debian unstable targetted at the next
 stable release.

> so, is this approach welcome, or do i waste my time by duplicating
> efforts (means: do any backport-policy vendors already have some kind of
> tracker systems to identify and backport serious patches)?

 Well, I do think there's room for improvement to save us all some time.
 I can imagine a score system similar to the "developer" points could
 help filtering important bugs which warrant a backport, but this would
 require some work; an alternative would be to use some "backport"
 keyword instead.  Useful information (as pointed in this list) would be
 the versions where the fix is present (2.16.3, 2.17.99 etc.).

 Thanks again for your efforts!

   Bye,
-- 
Loïc Minier <lool dooz org>