On Monday 10 September 2007 08:35, James Harper wrote: > > I'm still open on this, but I wouldn't release an official patch > > without > > > having it properly tested, so putting out an untested patch is not > > what I > > > consider a good solution. Even the current patches are not > > "officially" > > > released but are attached to an open bug report. > > Understood. > > > Can you explain why a patch would help with a "distribution packaged > > version of Bacula"? > > > >From Debian (the only distribution I'm really familiar with): > >From Debian (the only distribution I'm really familiar with): > > Sarge (oldstable) - 1.36.2 > Sarge-backports - 1.38.11 > Etch (stable) - 1.38.11 > Lenny (testing) - 2.0.3 > Sid (unstable) - 2.2.0 > > Do you believe that the bug is bad enough that any distribution should > not be releasing a version of Bacula that isn't patched? If that's the > case, then should Debian remove Bacula from Etch if it can not or will > not be patched? My understanding of the way Debian is works is that > updating Etch to 2.2.x just won't happen - bugfixes and security fixes > only.
I would recommend that Debian pull both 2.0.3 and 2.2.0. I am not familiar enough with their rules to comment on 1.38.11. Once a stable patch is available for one or both of those versions or a new release is available, they should proceed using their normal procedures. > > The 2.0.3 package from Lenny builds under Etch just fine. The 2.2.0 > package builds under Etch once you backport a few other packages, but > they are only going to get further and further apart, so in 6 months > time, someone wanting to backport the current version of Bacula might > have a bit of a problem. > > I have no problem with 'abandoning' the 1.x series from a bugfix point > of view (and I understand you aren't completely sure if the bug exists > in that stream anyway), With minimal testing on 1.38, I could not reproduce the problem, but I believe the bug exists in that version too. It would be a big amount of work to fix it on 1.38 and quite invasive (i.e. probably unstable). > but I think enough distributions have the 2.0.x > series that it would be worth the effort - or maybe they have different > release structures to Debian... > > Maybe the other alternative would be to carefully document the exact > situations which will and will not cause the bug to arise. IMO, I have already done a pretty decent job of doing that. Please read my announcement for the details. Once I have a stable test release (pre-production release) build and available, I will begin more detailed technical documentation of the problem and manual methods for restoring from existing backups. So far, no one but myself has confirmed that the patch works, so it is still too early to officially release it. > > As someone else pointed out, it would be a shame to damage the good name > of Bacula with something like this. > > That's my opinion anyway, thanks for listening :) > > James ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2005. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
