Looks like I manged to hit a bug in bash itself. It was triggered by repoman, let's see how upstream will handle it.

I just wrote a short script (named after cvsuck that can download a whole CVS repo with anon access only) that can download a whole picasa album without installing any firefox extension, binary software from Google, etc.

It's available here.

I'm just talking about 1.5+, since before that version it's like there would be not git merge, just git cherry-pick. (You had to explicitly specify each revision. That's not merging.)

A quick howto is here.

A command which may be useful but left out is like 'git log master..branch', to see what revisions will be merged:

$ time svn mergeinfo https://swig.svn.sourceforge.net/svnroot/swig/trunk/ \
--show-revs eligible
(...)
real    0m14.902s
user    0m0.155s
sys     0m0.013s

Well, it's getting better, but as the blog post says it's still not git-like.

Actually it's horribly slow, I'm still waiting for 'svn merge https://swig.svn.sourceforge.net/svnroot/swig/trunk/' to finish. ;-)

So when trying to understand others' code, and they use large while/for/if blocks, it's hard to see where one if ends, unless it's on the same page so vim highlights it. I remembered already once that there is a char to jump to the start { of such a block from its end } and vica-versa, but I forgot it, so now I had to look it up again. It's '%'. :-)

So my problem was that I already set up PGP, and I use it for sending / receiving signed (and sometimes crypted) messages. Today I received an S/MIME signed message and I had to verify it.

First mutt just bailed out with "Can't verify S/MIME signature", but that was because S/MIME was not configured. Here is what I did:

cd ~/.mutt
cp /usr/share/doc/mutt-ng/samples/smime.rc .
# edit muttrc to source it

I thought that will be enough, but that was too aggressive.

First, it disabled my PGP support. I had to put 'unset smime_is_default' to the top of smime.rc.

Second, it tried to verify the whole damn cert chain, which is simply something I can't really care, as the cert is a self-signed cert anyway. To do this, I had to uncomment the last two lines in smime.rc (set smime_verify_command/smime_verify_opaque_command).

So far it works just like I wanted it! :)

I wasted some hours debugging till I found out how to set the fontpath of a freenx-server if not using XFS.

Basically just need to fine-tune /etc/nxserver/node.conf:

DEFAULT_FONT_PATH=$(echo $(grep -v '^ *#' /etc/X11/xorg.conf | grep FontPath | awk \
'{print $2}' | sed -r "s/^[\"']|[\"']$//g") | sed 's/ /,/g')
AGENT_EXTRA_OPTIONS_X="-fp $DEFAULT_FONT_PATH"

Based on this post.

I just noticed that scootertechno.com has no RSS, so I wrote one. Available here, or in "binary form", here.

So swig uses two spaces instead of tabs, horrible. But I'll get used to it - just I don't want to fuck up my vimrc just because of their idiotic settings.

Here is the trick:

au BufRead */swig*/* set tabstop=8 | set shiftwidth=2 | set softtabstop=2 | set expandtab " SWIG
au BufRead */swig*/Source/* set tabstop=8 | set shiftwidth=2 | set softtabstop=2 | set noexpandtab " SWIG Source
au BufNewFile,BufRead *.i set filetype=swig
au BufNewFile,BufRead *.swg set filetype=swig

The swig syntax file is here.

So upstream recently released hal 0.5.12 and it's maintained by crazy who seem to be a bit busy currently, and I don't really wanted to bump his pkg (actually I tried and I failed), but I bumped our udev pkg yesterday so we no longer has libvolume_id. So I backported the relevant patch from git and it seems to work. ;)

(Actually the header is informative, but fake as I created the patch from scratch, based on the one which was in git.)

Okay, this is a crazy idea, but I love asciidoc. So the pipeline is the following: asciidoc -> docbook -> latex -> pdf -> odp -> ppt.

The fun part is that docbook is XML and then later I convert the stuff back to XML again (odp), but that's necessary as latex adds the nice themes and layout to the text.

Also the method is semi-automated as I generate latex from asciidoc, then copy&paste it to a latex-beamer document with some manual fixup, then I can convert the rest in a more or less automated way. Time to package odfpy and other affected, but not yet packaged components. ;-)