[please respond to debian-i18n ; I'm not on debian-devel]
Hello all !
I'm pretty proud to annouce that I've done my first valuable perl script :
it check in all source package you give him the status of i18n and l10n. It's
pretty alpha software (not packaged, no manpage, almost no args on cmd line),
but it's still usable.
My biggest problem is that I don't have a potato mirror. So, the only stats I
can do are on slink. That's why I'm looking for someone who have a good mirror
and wants to run it to see the result. The script is clue enought to not run
two times on the same package, so you can run it every day, it shouldn't take
too much time (the second time :).
But aware, it is a VERY big process, with very much disk acess, so it can
freeze your box on some buggy kernels (I've experienced this on 2.2.9). Use at
your one risk.
I've attached two tgz files. One is a buddle of scripts. unpack it in a
directory, install the liblocale-codes-perl from your prefered archive, run
'di18n-check' (with the path to the mirror as argument), and when it is
finished (more than one hour later on my box), run 'di18n2html'. You got the
files I want (first of all di18n.data)
In the other tgz file, you can find the stats of slink. I just put here the
ranking between languages (because I'm french :)
Rank Language Score #_of_pkgs average_l10n
1 French 12.8 44 87%
2 German 10.8 41 79%
3 Spanish 8.9 29 92%
4 Portuguese 6.2 23 81%
5 Italian 5.4 21 77%
6 Korean 4.8 19 76%
7 Dutch 4.4 17 78%
Polish 4.4 14 95%
8 Swedish 4.1 14 89%
9 Norwegian 3.7 15 74%
10 Czech 2.8 13 66%
11 Finnish 1.9 8 71%
12 English 1.7 6 89% (consider this as a bug :(
Russian 1.7 7 74%
13 Slovenian 1.0 6 52%
14 Croatian 0.9 3 99%
Turkish 0.9 3 98%
15 Irish 0.6 9 22%
Slovak 0.6 2 100%
16 Danish 0.5 4 39%
Japanese 0.5 4 44%
17 Greek 0.3 1 100%
Esperanto 0.3 1 100%
Galician 0.3 1 100%
18 Hungarian 0.2 1 74%
Icelandic 0.2 1 70%
Romanian 0.2 1 83%
Ukrainian 0.2 1 89%
19 Catalan 0.1 1 57%
Before I get flamed, I assume I have to say that there are partial stats :
- it's based only on po files (I can't stat catalogs for catgets()).
- only the po files of which I managed to guess the language are counted.
The score is calculed so : 20*(the average l10n)*(nb of packages l10n in this
language)/(nb of packages containing po files).
There is 60 packages containing po files in slink.
Ok, and now ? I plan to correct above noticed problems (args parsing, man
pages...) and package it. But I didn't manage to wait before I give it to the
world...
Bye, Mt.
PS: sory for the poor english, i'm VERY tired
Attachment:
script.tgz
Description: GNU Unix tar archive
Attachment:
slink.html.tgz
Description: slink.html.tgz