Hi, "Davide G. M. Salvetti" <[EMAIL PROTECTED]> (20/01/2007): > fant> Thomas Huriaux <[EMAIL PROTECTED]> wrote: > >> The main problem with this bug is that it forces translators to > >> understand your build system if they want to check their PO files. > > I'd like to know what exactly do translators need to check their new PO > files. Please, Thomas, I'd ask you to describe what you usually do when > you add a new po file to a package and what you would like the build > system to do for you. While you answer, please, do not assume that > every package maintainer should act the way most of them do, just state > what's your customary way to work as a translator (if it helps you, give > me a list of actions, like: I copy the "templates.pot" file to > "<LANG>.po", translate it, then a issue a "debian/rules whatever" while > pouring my cup of tee, ..., and so on, and so forth).
Here is what I do: - fetch the source package with apt-get source <package> - run debconf-updatepo to be sure to work on an up-to-date templates.pot - copy templates.pot into fr.po - translate fr.po - run debconf-updatepo to be sure that I haven't modified accidentaly a msgid - send the fr.po for proofreading on debian-l10n-french (this process is around one week) - apply the patches received - once this process is finished, I check if there has been new versions of the package during the proofread process, if so, I download the new source package and run debconf-updatepo to see if the file is still up-to-date. I usually even do it without checking if there has been new versions. - send the file to the BTS So, as I run 3 times debconf-updatepo in a normal translation process, I'd like to not have to understand the build system of every package I translate :-) > >> course, it is not as problematic as I thought before. > >> The other minor problem is that it gives false-positives in our i18n > >> infrastructure :-) > >> (see http://www.debian.org/intl/l10n/po-debconf/errors-by-pkg) > > I think you should follow lintian example and allow maintainers to > override the result of your checks if needed. Nobody knows better than > the maintainer (with the help of bug submitters) if the package is > right. I agree, but this is the first false-positive I've found. And it's even not a full false-positive, as it is recommended to not extract the strings from a generated file. Thanks for your answer and for your work, -- Thomas Huriaux
signature.asc
Description: Digital signature
