Re: Please keep an eye on manpage encoding issues (especially in etch).

[Clytie Siddall <clytie@riverland.net.au>]

Thankyou for explaining the situation for some other languages, Nori. :)

On 12/11/2006, at 6:35 PM, Kobayashi Noritada wrote:

> > > > > Manpage encoding issues are seen for some packages and for some
> > > > > languages; some manpages are encoded in UTF-8 and unreadable in  
> > > > > any
> > > > > environment.
> > > >
> > > > Can we specify the encoding in the manpage text in some way?
> > >
> > > No, the input encoding is determined by the locale and input manpage
> > > path, which is hard-coded in man-db's src/encodings.c.
> >
> > :(
> >
> > Users of some languages, like mine, will need UTF8 anyway, and will
> > set their man conf correspondingly. In any case, UTF8 will be used
> > increasingly. It is, or should be, the standard encoding for
> > internationalization.
>
> Yes, exactly I think so.  I hear Fedora have forced manpages to be
> encoded in UTF-8 and Debian should do so in the future, or at least
> support UTF-8 manpages for languages that have used non-UTF8 ones in
> some way (e.g. by detecting input encodings, or by completely
> switching to UTF-8).

Yes, indeed. :)
> However, completely switching manpage encodings to UTF-8 will require
> both modification of man-db's encoding handling and conversion of all
> the old non-UTF8 manpages (e.g. EUC-JP ones used for Japanese, and
> latin-1 ones used for European languages), which will be a big
> transition and will be controversial.  Actually it was controversial
> at least in a thread[1] starting from my post in this April in a
> Japanese list.

I can see that this will affect some languages which had viable  
encoding options before UTF8. Users of my language have been greatly  
motivated to use UTF8 by the fact that our legacy encodings are  
really clumsy and difficult to use. Plus a complete dearth of  
translators meaning we have no mainstream translated manpages (and  
not much of anything else) anyway!

But I know there has been mainstream and ongoing documentation for  
some other languages. It will be quite a task converting those docs.  
Maybe, with manpages, this task could be combined with reviewing  
manpages for currency. I know there is some concern over translated  
manpages which turn out not to be current, and in some cases are very  
old indeed.
> So, in the first mail, I proposed determining some policy about
> encodings of manpages and creating system to check manpage encodings
> in the future.  After that, or in parallel, we should do some
> transition or support.

Sound good to me. :)
> > But meanwhile, we need a way to label non-UTF8 manpages, or manpages
> > which don't match the man conf encoding setting. Bruno, what do you
>
> Yes, a number of non-UTF8 manpages are now used for many languages for
> which only non-UTF8 ones are approved.  And at least for etch, we now
> need conversion of UTF-8 manpages for those languages.
>
> [1] http://lists.debian.or.jp/debian-users/200604/ 
> threads.html#00078 (in Japanese)

(My Japanese is terrible, but...)

Yes, that's a fairly strong discussion. And dancer has a point, that  
we have to be careful about changing anything the system, or the  
software using it, expects to be in a certain format.

But I think we can do it, in time. :)

from Clytie (vi-VN, Vietnamese free-software translation team / nhóm  
Việt hóa phần mềm tự do)
http://groups-beta.google.com/group/vi-VN

Attachment: PGP.sig
Description: This is a digitally signed message part