September 2009 - Arch-dev-public - lists.archlinux.org

[arch-dev-public] [signoff] kernel26-lts 2.6.27.31-2
by Andreas Radke 05 Sep '09

05 Sep '09

This is our new Long Time Supported Kernel recommended for server usage and our general fallback kernel that already made it into the distrowatch news section :) It should now be ready to move into the repos. Please install it along the kernel26 and add it to your grub/lilo section. Give it a try but don't expect any 3rd party modules or late grafics stuff for desktops. Only full Ext4 support has been added for fallback reasons. Please report if you can boot well and if dmesg is clean. Choice of repo. Most devs say it belongs to core. I'm fine with that. Is it safe to push it to core? Some people have concerns about the iso image size. -Andy

3 3

[arch-dev-public] [signoff] man-db 2.5.6-1
by Andreas Radke 05 Sep '09

05 Sep '09

new upstream release. please test it and signoff. -Andy About this release ------------------ This is largely a bug-fix release, but also includes a new 'man -K' option to search the full text of all manual pages, as well as support for interpreting pairs of manual page names given on the command line as equivalent to a single manual page name containing a hyphen (e.g. 'man foo bar' => foo-bar(1)) suggested by various posters to the git mailing list. For full details of these and of the other improvements in this release, please see the files docs/NEWS and docs/ChangeLog in the source distribution. man-db 2.5.6 (26 August 2009) ============================= Major changes since man-db 2.5.5: Fixes: ------ o Exact-section database lookups were incorrectly returning all database entries whose section names were prefixes of the requested section name. In some cases this could confuse mandb into never believing that the database was up to date. o Fix handling of pages with comma-separated names ("foo, bar, baz") in their NAME sections, broken by a change in 2.5.0 (!) to ignore manual page names containing spaces. o Fixed a buffer overflow in the pipeline library's line-reading functions. I don't believe this to be exploitable: at worst we might believe that there's some garbage at the end of manual pages (whose contents are untrusted anyway) and this bug typically resulted in a failed assertion the next time anything tried to read a line. o Plugged two substantial memory leaks in the pipeline library. o whatis and apropos only display any given manual page, or pointers to it, once. o man now sets less(1)'s environment up correctly for manual pages encoded in CP1251. o manconv no longer confuses situations such as "this UTF-8 character is not representable in the target encoding" with "this text is not in UTF-8". Improvements: ------------- o The default configuration file now includes section 0, used on some systems to document C library header files. o 'make check' now passes in the presence of a UTF-8-aware col, such as that in util-linux-ng. o The 'man -K' option is now supported to search the full text of all manual pages. This was inspired by a similar option in the other man package (currently at version 1.6f) currently maintained by Federico Lucifredi and formerly by Andries Brouwer, but I took advantage of man-db's pipeline library to implement it entirely in-process, without having to start a separate grep process for every manual page. In my tests with fairly typical searches across variously all manual pages or just one section, man-db's implementation ran between 3 and 10 times faster. o Database directories are now only created when there are corresponding manual page directories, not just because they're mentioned in the configuration file. o By default, man will now try to interpret pairs of manual page names given on the command line as equivalent to a single manual page name containing a hyphen (e.g. 'man foo bar' => foo-bar(1)). This supports the common pattern of programs that implement a number of subcommands, allowing them to provide manual pages for each that can be accessed using similar syntax as would be used to invoke the subcommands themselves. Suggested by H. Peter Anvin, Federico Lucifredi, and others on the git mailing list. o The build process is now quieter by default. Use './configure --disable-silent-rules' or 'make V=0' if you don't like this or your make(1) doesn't support the non-standard extension required. o 'make install' now installs the manual. o manconv understands a wider range of Emacs-style coding tags. o Recommendations to change MAN_DB_CREATES, MAN_DB_UPDATES, and MAN_CATS #define options in manconfig.h have been replaced by new configure options --enable-automatic-create, --disable-automatic-update, and --disable-cats respectively. Note that automatic user database creation is now off by default, as it is often too slow for the usefulness it adds; use --enable-automatic-create to enable it.

3 2

[arch-dev-public] [signoff] cloog-ppl-0.15.7-1
by Allan McRae 05 Sep '09

05 Sep '09

Upstream bug fix release. To test, try out some of the graphite optimizations in gcc and if stuff works... Yeah, no-one is going sign-off this one! :) Anyway, it passes make check on both arches. Signoff both. Allan

2 2

Re: [arch-dev-public] vim status?
by Tobias Kieslich 05 Sep '09

05 Sep '09

Hi, It seems there are no concerns on ex-vi as vi replacment so I will get vim out of there in the next 2 days. I will rebuild the plugins tonight. -T On Wed, 12 Aug 2009, Allan McRae wrote: > Hi, > > What is the current status of vim in [testing]? I would like to > move the ruby-1.9 builds to [extra] at some stage but this requires > the vims in [testing] to move. > > Allan > >

2 1

[arch-dev-public] Backpacking this weekend
by Aaron Griffin 04 Sep '09

04 Sep '09

I'll be away from Saturday - Monday on a backpacking trip in Southern Illinois. I'll have my cell for email, but who knows if it will work. If you really need me, I'd recommend getting a hold of Dan and having him either do what needs doing, or sending me a text :)

2 1

[arch-dev-public] Moving DB scripts live
by Aaron Griffin 04 Sep '09

04 Sep '09

So, we have "any" support and split package support sitting in the DB scripts in /arch-new I'd like to move this live soon. What do you all think? Can you test normal packaging with these to confirm there's no major regressions?

5 12

[arch-dev-public] permission to adopt package in communtiy repo
by Andrea Scarpino 03 Sep '09

03 Sep '09

Hi, I am still a TU so please give me permission on dev platform to adopt my packages in [community] repository. Thanks -- Andrea `bash` Scarpino Arch Linux Developer

2 1

[arch-dev-public] German Pronunciation
by Aaron Griffin 03 Sep '09

03 Sep '09

How do you say "Gudrun". "Gerolde", and "Sigurd". I have my own English-ized pronunciations, but I'd like to know the real way to say it. I imagine the "Gud" is kinda like the german "guten", but is the "run" part like that too? Gude-rune?

3 2

[arch-dev-public] Reducing traffic usage during package move between repos
by Roman Kyrylych 02 Sep '09

02 Sep '09

On Wed, Sep 2, 2009 at 11:41, Dieter Plaetinck<dieter(a)plaetinck.be> wrote: > I think a big improvemt will be when we fix it so that if we move packages from testing to extra/core, mirrors will not see as a "file remove and a new file", which causes a lot of unneeded traffic. > if we make this system use symlinks or whatever this can be handled much better. > but other then that, I think the current syncing system is fine. (the "getting" part is something different. see tiered setup,mirrorbrain etc. discussions) Since we already use symlinks for -any packages this proposal looks okay. The layout could be like this then: <repo>/os/i686/foobar-1.2.3-1-i686.pkg.tar.gz -> ../all/foobar-1.2.3-1-i686.pkg.tar.gz <repo>/os/x86_64/foobar-1.2.3-1-x86_64.pkg.tar.gz -> ../all/foobar-1.2.3-1-x86_64.pkg.tar.gz <repo>/os/any/foobaz-4.5.6-1-any.pkg.tar.gz -> ../all/foobaz-4.5.6-1-any.pkg.tar.gz <repo>/os/all/foobar-1.2.3-1-i686.pkg.tar.gz -> ../all/foobar-1.2.3-1-i686.pkg.tar.gz <repo>/os/all/foobar-1.2.3-1-x86_64.pkg.tar.gz -> ../all/foobar-1.2.3-1-x86_64.pkg.tar.gz <repo>/os/all/foobaz-4.5.6-1-any.pkg.tar.gz -> ../all/foobaz-4.5.6-1-any.pkg.tar.gz As mentioned above - this will reduce the amount of traffic during package moves between repos. The negative side is that this will need one-time redownload of all packages by mirrors, though this can be avoided by manual moving of the files by mirror admin (need to be done in this way: wait until /all syncing starts, kill rsync, move packages manually, start rsync). BTW I always wondered why we have that silly "os" subdir. Would be nice to get rid of it during this big mirroring change (if it'll happen). -- Roman Kyrylych (Роман Кирилич)

4 6

[arch-dev-public] subversion older in testing
by Andreas Radke 02 Sep '09

02 Sep '09

we have svn pkg 1.6.4-2 in testing and a later 1.6.5-2 in extra-x86_64 and an older one in extra-i686. any reason? I assume we need to rebuild 1.6.5-2 for i686 and clean up the ones in testing? -Andy

5 6