just finished another regression test for the testsuite: srcmissing, which searches for missing (not yet mirrored) source tarballs (and optionally downloads them). available here

in the last two days i was working on creating an initial testsuite for the source tree. what does works:

• cleaning outdated fpms
• cleaning outdated sources

this is the harder part imho what is planned:

• list missing fpms (maybe not needed? checkpkgs does the same)
• download all missing sources
• check for syntax errors
• check for required directives

i always was lame to the advanced usage of chanserv - and now again, i had to do something. the task was simple: set up autovoice'ing for a channel at freenode

so the two magic command:

/msg ChanServ LEVEL #chan SET autovoice 1

and then for each user:

/msg ChanServ ACCESS #chan ADD username 1

that's all ;)

just a 2 cheap trick with vim for other C monkeys:

:map  :set makeprg=make:w:make
:map  :set makeprg=make\ test:w:make

(probably these explain themselves)

the other nice feature: move to a funcion and press K: that will show the manpage of the function

hm. this was easier than i tought. so. it's fwcpan

$ fwcpan inline::cpp
Searching CPAN... done.
Guessing module: Inline::CPP
Targets: perl-inline-cpp-0.25-1
Proceed with upgrade? [Y/n]
Writing the FrugalBuild... done.
$ cat FrugalBuild
# Last Modified: Wed, 08 Feb 2006 03:04:56 +0100
# Compiling Time: 0.1 SBU
# Maintainer: Zsolt Szalai <xbit@frugalware.org>
pkgname=perl-inline-cpp
modname=Inline-CPP
modauthor=N/NE/NEILW
pkgver=0.25
pkgrel=1
pkgdesc="Write Perl subroutines and classes in C++."
depends=('perl')
groups=('devel-extra')
archs=('i686')
Finclude perl
sha1sums=('1dd7bdbe1a14f3c824841c72a69c54b07dbe9575')

i've pushed it to pacman-tools, it'll be included in the next release :)

openoffice dictionaries are back. a new openoffice-dict scheme was created, with current 4 dicts: de, fr, hu and es

there are new setup snapshots out. no big changes, just the static pacman from the setup has gone and the select plugin is now using libalpm calls instead of system(). they're available at the usual place: i686, x86_64

i'm working in decreasting the package size of the openoffice.org package with subpkgs() 195M 2005-12-24 13:31 openoffice.org-2.0.1-1-i686.fpm ^ this was the original one 83M Feb 5 13:20 openoffice.org-2.0.1-2-i686.fpm it seems that i'm on the way :)

today i got this "having subpackages in extra while the main pkg is in the frugalware repo" stuff working. a few notes from the manpage: Special case: subpkgs in other repos.

The situation is the following: you have a package which is in the frugalware repo, but you want its subpackage to be in extra. It's a but tricky, but still possible.

The trick is to create a fake FrugalBuild in the other repo. You could place it whereever you want, but to use repoman you need to place it under $fst/extra/source/subpkgs/pkgname, where pkgname is the name of the original main package.

How to create that fake FrugalBuild?

Copy the original one, but remove everything except the subfoo directives, pkgver and pkgrel. The next step is to clarify which subpackage have to be in which repo. This is done by removing the architectures from the subarchs() array.

Example:

The old main:

subpkgs=('foo' 'bar') subarchs=('i686 x86_64' 'i686 x86_64')

The new main: subpkgs=('foo' 'bar') subarchs=('i686 x86_64' '')

The new fake: subpkgs=('foo' 'bar') subarchs=('' 'i686 x86_64')

This means that bar is expected to be uploaded to extra, while foo is still in the frugal- ware repo. If you move all the subpackages to extra, then of course you can remove the whole subarchs() array from the main FrugalBuild. a few other notes:

• it requires repoman from pacman-tools-0.6.3
• subarchs() is no longer optional
• as the fake FB is not in the current dir, if you record with repoman, it won't record the it!
• i've added subarchs() to the existing packages and updated the skel file :)
• for database generation it needs pacman from darcs, but probably you don't have to care about it as genesis runs that version

after a few fixes now it seems that syncpkgd really does what it should. now it runs on factory (i686) and helicon (x86_64)