[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Deactivated languages



Pe data de Lun 20 Noi 2006 16:25, Jens Seidel a scris:
> Hi Eddy,

Hello Jens,

> > Actually, not quite. The biggest problem is the size of the initrd image
> > on which there must be translations for countries, languages, and also
> > all the translation for all the activated languages and all the packages
> > in the initrd.
>
> I'm aware that the size is a problem. But it was never mentioned that
> the size is the reason for dropping a language. IIRC there was never a
> fully translated language dropped recently because of size restraints
> (I know about the old Vietnamese situation for the Sarge release).

I guess dropping a full active translation would be counted as 
unsocial/rude/bad-taste/whatever by anybody, that's why it has never been 
seen and will never be.

Actually, the initrd issue was present since the start, and since Christian is 
so good at mobilising people to translate D-I, and the problem was more and 
more visible, more and more a real solution was wanted. Until now it was 
never decided "for good" how to deal with this and the issue was postponed in 
the hope it will be solved somehow. Of course, this lead to all kinds of 
undesirable compromises like arbitrary translation ratios, translation 
activity - (please do not start the issue again) - following the path of 
least opposition. All of these are what *I* know/think about the issue.

> Reading a private mail from Holger (the German manual translator) I got
> the idea to omit rarely used message in de.po (and maybe other
> languages, depending on the usage?) in favour of dropped languages to
> avoid larger images.

This, like many other solutions, is suboptimal since an expert installation 
will hit those. The real solution would be to actually address the issue 
without loosing/dropping translations and keeping the initrd size at a small 
size.

I would be really curious to find out the percentage of taken space by the 
translations in the initrd image, as it is now, and how does a new complete 
translation impact on the size of the image.

> > During the writing of this mail (offline) it hit me: I wonder if a list
> > with all the countries each in its own language/languages wouldn't reduce
> > considerably the size of the image? It would be a bitch because of the
> > Han unification, but we already do it (somewhat in a correct fashion) in
> > G-I. Christian, didn't we discussed about this in September?
>
> That's not easy to understand but I assume you want to ensure that not
> each language specific font has to use glyphs which are only used in the
> language list? You propose to use only one font for all languages ro
> display the countries. Is this correct or did I misunderstood you?

I guess I wasn't to clear. This doesn't have to do with glyphs and fonts.

I was suggesting that the country list could have only translation in the 
native language(s) spoken in that country. This could be a nice study to do 
to analyse the translation/image size relation.

This small "optimisation" would start from the assumptions:
- someone choosing language A spoken in countries X,Y and Z would see only the 
country names X,Y and Z translated in language A
- if you live in country W you are likely to know (at least) the name of the 
country as written in at least one of the (official) languages spoken in W, 
so if you choose language B (not official in W) you will be presented with 
list of countries in which B is official (exactly like now), but if you 
choose "Other" a list with all countries in the official language(s) present 
there (something like in the example below):

Deutschland
מְדִינַת יִשְׂרָאֵל
Italia
România
Росси́я
United Kingdom

> > As I said, the problem is the actual space on the initrd and having a
> > "latin languages image/CD" and a "asian languages image" is not such a
> > good idea since you might end up downloading the wrong image and will be
> > faced with a semi-useless cd  since you don't know any of the other
> > languages.
>
> I fully agree.

Thus, we must find real solutions, or some optimisations

> Yep, but it's hard if the problems are explicetely hidden. Now I'm aware
> of the size contraints and a missing fallback language.

Note that the support needs to be added in Debconf.

> According to my understanding of the issue the problem where not
> necessarily technical issues.

Well, it is :-)

-- 
Regards,
EddyP
=============================================
"Imagination is more important than knowledge" A.Einstein



Reply to: