[pacman-dev] Thoughts of a 3.1.2 release
Here is the current git shortlog since the last release: $ git shortlog v3.1.1..maint Alex Merry (1): updatesync: incorrect package deletion logic Chantry Xavier (9): workaround for a gettext string starting with --. two string fixes. makepkg : improve determination of svn revision. Clarify the "cancel current operation" message. repo-remove: print an error in case of an empty db Clarify the "failed to add target" errors. add 'force' to PKGINFO, have repo-add respect it. Add new ChangeLog.proto file. Don't follow symlinks with -Qo. Dan McGee (7): makepkg: add check to ensure we have non-URL files in build dir Remove frontend translation of "debug:" message doc/makepkg.8: add missing '-' to manpage option makepkg: ensure binaries in /opt/* are stripped Add deprecation warnings to gensync and updatesync doc/makepkg: document --log, address locale issues pacman/util.c: add mdirname function Roman Kyrylych (1): Remove /bin/true from install.proto Given this list of changes, I think it would be prudent to think about a 3.1.2 release soon, although it is not something we need to rush out as there is nothing pressing like the provides logic prior to 3.1.1. Here are some issues I think we should address before prepping for release: http://bugs.archlinux.org/task/9362 - Any opinions here? Put them in the bug report. http://bugs.archlinux.org/task/9242 - Still not completely sure if this is easily solvable http://bugs.archlinux.org/task/9230 - Deals with makepkg/versionpkg stuff, should be an easy fix Please add any (constructive) comments you may have, and quote the relevant parts of this email. *Please* keep it limited to things that would make it into 3.1.2- if it is clearly something that can't fit in a maint release, I won't even bother responding. A second email will follow this one with some things that Xavier has brought up he thinks need addressing. Some of these are relevant for 3.1.2, others may be a bit further in the future. -Dan
Seriously guys? No replies? On Feb 5, 2008 8:40 PM, Dan McGee <dpmcgee@gmail.com> wrote:
Here is the current git shortlog since the last release: ... Given this list of changes, I think it would be prudent to think about a 3.1.2 release soon, although it is not something we need to rush out as there is nothing pressing like the provides logic prior to 3.1.1.
Here are some issues I think we should address before prepping for release: http://bugs.archlinux.org/task/9362 - Any opinions here? Put them in the bug report. I'm leaning towards "Won't fix" here.
http://bugs.archlinux.org/task/9242 - Still not completely sure if this is easily solvable Not sure here. Doesn't seem like a release blocker though.
http://bugs.archlinux.org/task/9230 - Deals with makepkg/versionpkg stuff, should be an easy fix Fixed and checked in.
FS#9235 was also fixed as well tonight- thanks Xavier.
Please add any (constructive) comments you may have, and quote the relevant parts of this email. *Please* keep it limited to things that would make it into 3.1.2- if it is clearly something that can't fit in a maint release, I won't even bother responding.
A second email will follow this one with some things that Xavier has brought up he thinks need addressing. Some of these are relevant for 3.1.2, others may be a bit further in the future.
Above comments still apply here. Please get all things under consideration for 3.1.2 into this thread. String freeze is coming soon, so be ready translators. Feel free to get a head start as well. Hopefully Giovanni can manage this for me once again once we are ready. :) -Dan
Dan McGee wrote:
Here are some issues I think we should address before prepping for release: http://bugs.archlinux.org/task/9362 - Any opinions here? Put them in the bug report.
I'm leaning towards "Won't fix" here.
http://bugs.archlinux.org/task/9242 - Still not completely sure if this is easily solvable
Not sure here. Doesn't seem like a release blocker though.
I think these are really the same issue. makepkg extracts files with the system umask, which if not sane can cause problems if file are then installed poorly. I've added a comment in FS#9242 showing that I think the opera install.sh uses "cp" to install files. The only justification I can think of to fix this is if for some reason extracting with such permissions causes a package not to actually build. Then we should probably set the umask earlier. Can this happen other than with a really stupid umask? Otherwise, a check of the current umask could be added and a warning printed if not 0022. Allan
Allan McRae wrote:
I think these are really the same issue. makepkg extracts files with the system umask, which if not sane can cause problems if file are then installed poorly. I've added a comment in FS#9242 showing that I think the opera install.sh uses "cp" to install files.
The only justification I can think of to fix this is if for some reason extracting with such permissions causes a package not to actually build. Then we should probably set the umask earlier. Can this happen other than with a really stupid umask? Otherwise, a check of the current umask could be added and a warning printed if not 0022.
Ah yes, indeed, that's it. I couldn't figure that out alone :) Well, attaching a patch to set the umask earlier. That seems to work fine.
participants (3)
-
Allan McRae
-
Dan McGee
-
Xavier