Re: New translations

From: Chris Leonard <cjlhomeaddress_at_gmail.com>
Date: Sat Mar 24 2012 - 02:43:59 CET

On Fri, Mar 23, 2012 at 9:11 PM, Simon Larochelle
<larochelle.simon.1@gmail.com> wrote:
> Second try: I have a patch to generate the '.strings' files with make.
> A few things:
>
> 1) The '.strings' files would be created in the 'po' directory instead
> of in the 'user/wp/strings' directory because of a limitation of
> automake.
> 2) As the '.strings' files are generated with a perl script, we would
> need to add perl to the list of required programs in the configure.in
> file.
> 3) Presently, two of the po files are not installed with 'make
> install': mnk-SN.po and ps.po. Should the files be added to the
> installation script or are they incomplete?
>
> Simon

Simon,

Mandinka is not 100% complete, but at 95% for 2.8 and 94% for 2.9 it
is certainly worth including.

http://translate.sugarlabs.org/mnk/upstream_POT/

There are a few printf errors in lang-mnk (which in my Sugar
experience can be build breakers), but I have made them "fuzzy" in the
Pootle version. This is only speculation, but if lang-mnk was
excluded for causing build problems, fuzzy-ing these printf error
containing strings might temporarily mask any build issues. Thatt is
a trick I've used successfully at Sugar Labs to help our release
managers move forward when blocked by L10n-related fatal errors during
a build.

Pashto is a little less complete at 73% for 2.8 and 73% for 2.9, but
IMHO that is still a substantial and very usable amount of L10n
coverage.

http://translate.sugarlabs.org/ps/upstream_POT/

I only took a quick look at the pofilter flags on Pootle, but I didn't
see any obvious reasons that these PO files would be problematic
(printf and terminal newline mismatch being the most common
build-braking L10n errors), so I can't even speculate as to why it
might be intentionally excluded, probably an oversight. Failing to
update a language list kept separately is unfortunately a L10n
management error I have made myself.

Thank you for exploring the possibilities automating elements of the
L10n process and the potential that creates for disintermediating
human L10n management speedbumps.

cjl
Received on Sat Mar 24 02:44:49 2012

This archive was generated by hypermail 2.1.8 : Sat Mar 24 2012 - 02:44:49 CET