Joey Hess
2006-Mar-13 12:28 UTC
[Secure-testing-team] Guidelines for testing security fixes
Moritz Muehlenhoff wrote:> Quoting secure-testing.debian.net: > 1. Only upload changes that have already been made in unstable and are > blocked by reaching testing by some other issues. This is both to keep > things in sync once the new version from unstable reaches testing, and > to avoid breaking secure-testing too badly with fixes that have not been > tested first in unstable. > > Rebuilding packages from sid into testing that are stuck by a larger > transition does not work if the maintainer has fixed a vulnerability > by upgrading to a major new upstream version. E.g. kate: > testing has 3.3.2 and the sid fix is kdebase 3.4.1. In cases like > this we should better prepare patched packages for 3.3 (in the > case of KDE it''s rather simple as there are official upstream > patches for this version). As long as the packages in sid are fixed > as well, things don''t become out of sync here and official upstream > patches should normally be tested sufficiently. > > Comments?Right. On the other hand patching 3.3 before unstable has any fix for the problem could cause issues once unstable got a new version (possibly w/o the fix) and is the situation that guideline is intended to prevent. Backporting fixes is of course OK. Perhaps s/changes/fixes/ on the first line would clarify it. -- see shy jo -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 189 bytes Desc: Digital signature Url : http://lists.alioth.debian.org/pipermail/secure-testing-team/attachments/20050829/114bd92a/attachment.pgp
Moritz Muehlenhoff
2006-Mar-13 12:28 UTC
[Secure-testing-team] Guidelines for testing security fixes
Hi, Quoting secure-testing.debian.net: 1. Only upload changes that have already been made in unstable and are blocked by reaching testing by some other issues. This is both to keep things in sync once the new version from unstable reaches testing, and to avoid breaking secure-testing too badly with fixes that have not been tested first in unstable. Rebuilding packages from sid into testing that are stuck by a larger transition does not work if the maintainer has fixed a vulnerability by upgrading to a major new upstream version. E.g. kate: testing has 3.3.2 and the sid fix is kdebase 3.4.1. In cases like this we should better prepare patched packages for 3.3 (in the case of KDE it''s rather simple as there are official upstream patches for this version). As long as the packages in sid are fixed as well, things don''t become out of sync here and official upstream patches should normally be tested sufficiently. Comments? Cheers, Moritz