Re: non-utf8 po files breaking the build

[Kjartan Maraas <kmaraas online no>]

fre, 2002-12-06 kl. 01:57 skrev Keld Jørn Simonsen:
> On Tue, Dec 03, 2002 at 01:43:26AM +0100, Christian Rose wrote:
> > tis 2002-12-03 klockan 00.42 skrev Keld Jørn Simonsen:

[SNIP]

> > 
> > Part of the purpose of the latest messages in the most recent thread was
> > to make it explicitly clear that we have decided to keep GNOME 2 po
> > files in UTF-8 in cvs, so that we wouldn't need to have this discussion
> > over and over, and some people always claiming things aren't decided. It
> > seems we failed to make this clear to everyone.
> > 
> > You should have a look at
> > http://mail.gnome.org/archives/gnome-i18n/2002-November/msg00201.html : 
> > "The only thing that has been decided
> > is that all po files should be in UTF-8 format *in* *CVS* so that we
> > have less problems for the people trying to make tarballs once a week
> > and also for the people who are trying to keep the tree buildable at all
> > times."
> > 
> > If you read Kjartan's and my responses and many else's, it should come
> > as no surprise that there is a large consensus here.
> 
> I dont see a huge consensus. You and Yanko, yes, but then some people
> saying yes and no and but... And my questions on why were not answered.
> 

The question was answered several times. It's much easier for the
maintainers to have a single encoding to worry about than having all
different encodings in CVS. Scripts/tools don't have to handle every
local encoding out there etc.

> > It should also become clear that things are actually decided. I don't
> > like having to point this out, but Kjartan is the GTP coordinator and
> > has been so for quite some time. Kjartan and me are GTP contact persons.
> 
> Kjartan said we should try to make it so that the translators were not
> affected. I had a proposal to do this, which was not discssed at all.
> 
Nothing is stopping you or anyone else from fixing this right now. The
project has decided to keep stuff in UTF-8 in CVS to make it easier on
the maintainers and other developers, and I think that lessening the
burden on the maintainers is far more important than me having to run
iconv or any other conversion tool on the translation every time I check
it in. That is *if* I choose to use local encoding when doing the
translations.

> > > > Please remember that things like this don't get decided by popular vote,
> > > > and the decisions certainly don't get invalidated by one persons
> > > > reluctant protests on having to commit things in UTF-8, no matter how
> > > > loud the complaints/rants may be.
> > > 
> > > Oh well, but just having two loud ranting persons advocating UTF-8
> > > should not make a decision either:-)
> > 
> > Sometimes it depends on who those people are (also, there wasn't only
> > two people advocating UTF-8). We certainly don't want a situation where
> > things are decided by only a copuple of people, but we don't want a
> > situation where nothing gets decided and we can't move forward either.
> 
> It is fine that the project moves forward, but to me UTF-8 only
> is not moving forward, but shooting ourselves in the foot.
> 
Why?

> > That's why the projects have coordinators, and lately also contact
> > persons. That's why decisions can be made like for example in the GDP,
> > where they decided to move from SGML for all GNOME documentation to
> > Docbook XML. It may have been an impopular decision for some, but there
> > was a consensus that this was the right direction and it was eventually
> > decided.
> > That's why there are coordinators -- to represent the individual GNOME
> > subprojects. That means that there has to be some decisions in order to
> > represent the subproject as a whole, and to work with the same goals.
> 
> Well, who chose them?
> 
Nobody chose them. That means, coordinators, maintainers etc are there
because of the work they did in the area they coordinate or maintain. If
someone wants to it's just a matter of stepping up to the plate. Nobody
ever chose me as coordinator or translator for that matter. I did the
work and ended up being named coordinator after a while.

> > > I proposed a way that could accomodate everybody, I think,
> > > namely to extract and convert to utf-8 when you need the po-files,
> > > eg for statistics. I ahve not seen any comments on that.
> > 
> > You come with paper solutions; "maybe it can be done this way", "maybe
> > files can be automatically recoded", etc. I haven't seen any concrete
> > bug fix yet. Remember that code speaks more than words, and we haven't
> > seen any code yet from you to fix the current GNOME 2 i18n
> > infrastructure to work with multiple encodings.
> 
> Well, I could try to if you think it could solve the problem, 
> but you do not give me the chance at all.
> 
Don't give you a chance? Show us the code :-)

> > In any case, the use of UTF-8 for GNOME 2 po files in CVS has been
> > decided. Please refrain from more mails claiming the opposite.
> 
> I don't see it has been decided. There are proposals on the table
> that could make such a solution less desirable.
> 
For now it has been decided, and the po files have been converted. And
likely will be converted again if people start commiting other
encodings.

It's not like we're trying actively to place a burden on translators and
make their lives harder here. Maybe a couple of minutes of hacking in
CVSROOT to make the checkin scripts do the conversion is all that is
needed? Not sure how well that will work for all encodings though. The
script would need to know when and how to convert to/from all different
encodings to/from UTF-8, right?

Cheers
Kjartan