Re: File Formats and Names in Bug Reports
> I care about making things easier for all of us. I think standardization
> does so. It appears that you disagree.
I mostly think, as others, that standardization may help but will be
very difficult to achieve. The translators population is a very moving
target, most often people who want to help but are not highly
technically skilled (not all of them: some are also very highly
skilled...).
The translation team history is also different from one language to
another. For instance, the use of compression is widely developed in
non-Latin languages "teams" because they had experience with encoding
mess...
There can be some recommendations for file names. Kostas mentioned the
one we use in D-i, which is not bad.
However, this request that the translator indeed *knows* where the
file goes in the package source tree...which is not obvious when using
the tools that have been setup by some teams.
About the name of the file, and implicitly the language ISO code, for
sure some effort is needed. Translators should use a name which
includes the correct ISO code (indeed, I think that John mail comes
from a bug report I made about some PO file which was incorrectly
named in one of his packages).
Most of the time, what is sent by translators is quite easy to
understand. So we maybe just need to spread out the word about taking
care for naming the sent files.
John, about ISO codes, may I point you to the iso-codes packages,
which includes the ISO-639 (languages names and codes) tables. This
may help you in cases you have some doubt.
Of course, asking for help here or directly to the translator is also
always an option.
My own practice (and that of most French translators), is sending a
file named fr.po, not compressed, with a text explaining where the
file is (programs translation or debconf translation, most often).
Some hints about language codes:
-while they theoretically can be 3-letters codes, all translations
currently received for Debian packages, except very rare translations
coming from upstream (most of them in the iso-codes package), are for
languages with 2-letter codes
The 2-letter code should be used
-they should NOT be a country part to the translation file, EXCEPT
FOR:
pt_BR (Portuguese as spoken in Brazil)
zh_TW (rather than "Chinese as spoken in Taiwan", this is
"Traditional Chinese")
zh_CN (rather than "Chinese as spoken in China", this is
"Simplified Chinese")
All other xx_YY combinations should be renamed to xx, as of the
current state of Debian i18n. If in doubt, ask here.
Reply to: