Re: lynx-dev address@hidden: Re: new version of FR.PO][PATCH]

[Doug Kaufman]

On Wed, 22 Jan 2003, Henry Nelson wrote:

> My opinion is that it is the job of the person making the *.mo to make any
> and all necessary character set conversions.  If it isn't or can't be done
> at that level then the person installing Lynx will have to deal with it.

I was going to say that the person making the .mo files and the person
installing lynx are generally the same, but at least in the case of DOS,
the users are usually going to get a premade .mo file. There is no way
to know, however, which character set the user is going to have on their
machine. I switch mine regularly between CP437 and ISO-8859-1 (same as
CP819). I also switch to CP850 when testing the .mo files. Western European
users frequently use CP850 as their default. Eastern European users are
presumably using CP852 or ISO-8859-2. The same .mo file should work for
each system.

> Since DOS is not a multi-user OS, it seems preferable to do the character
> set conversion at the *.po level, according to individual needs, and then
> compile the *.mo file.  Done that way, libiconv is unnecessary.

Except that the DOS port of libintl is dependent on libiconv. It is
possible, but difficult to get libintl without the libiconv dependence.
  
> I strongly believe that it is not the job of the translator to have to worry
> about any of the technical issues beyond making a workable translation.  It

I agree. I think it should be an administrative job to make sure that
the translation has a valid character set in its header.

> People making *.mo files for a particular platform in a particular
> character set might offer these to the public if they wish by posting
> to lynx-dev the information on how to obtain it.

I think that this might be a major headache to maintain.

What is the current status of the internationalization effort for lynx
on platforms other than DOS which use different character sets. For
example, the Russian .po file is in koi8-r (or is that KOI8-R). How is
this handled in a lynx binary running on a platform with a display in
ISO-8859-5? Does this work with the .mo file generated by the usual
configure and make procedure of lynx, or will someone have to convert
the .po file to the display character set before building lynx? The
DOS port is built with libiconv so that this isn't a worry, as long
as a valid character set that is recognized by libiconv is in the .po
header.

One other thought. Lynx already has the character set translation code
built in for viewing documents written in character sets other than
the display character set. Should we think about extending that to the
strings brought in by gettext? Is that even possible with the gettext
mechanism?

                        Doug

-- 
Doug Kaufman
Internet: address@hidden


; To UNSUBSCRIBE: Send "unsubscribe lynx-dev" to address@hidden