Re: GNOME and migrating to utf-8

[Michael Meeks <michael ximian com>]

Hi Kjartan,

On 8 Mar 2001, Kjartan Maraas wrote:
> Recently some of the translators decided to move to utf-8 encoding for
> parts of their translations, which in it self is a step in the right
> direction. The problem as far as I know is that the current GNOME   
> codebase isn't prepared to handle utf-8 encoded strings in .desktop-,
> .oaf-, .glade-files etc.

        I suggest caution here; abandoning my own advice I will ramble on
a subject about which I know, oh, almost nothing - so here goes:

        Issues:

                * libxml1 is broken wrt. utf-8 [1]

                * libxml1 is used to parse oaf, and probably other XML
        files needing translation

                * We can't render UTF-8 strings correctly without Pango.
        
                * With my 10 minute research it seems that gettext doesn't
        allow you to determine whether a given translation is UTF-8 or
        some indestinguishable random 8bit string. So we can't convert    
        UTF-8 strings to some 8 bit subset to render, since we can't tell
        it's UTF-8.
        
        So, in short, I imagine that if you start using UTF-8 a lot,
things will get very badly broken, and we cannot detect or fix the
breakage in the code.

> The problem of GNOME not handling this is also visible in that we
> can't display the KDE menus correctly since these are in utf-8.  
> 
> We should probably wait until 2.0 before we migrate, but I'd like to
> start a discussion on this topic so that we can shed some light on all
> implications and how to attack this in the best possible way.

        When the mystic day of Gtk+2.0 release arrives, we will have to
get some small scripts written to UTF-8-ize the existing non UTF-8
documentation, and port it all across in a big conversion-fest. Until then
we need to stick with our whacked out mix / mess of encodings I imagine.

        Regards,
        
                Michael.

[1] - amazing, but true.

-- 
 mmeeks gnu org  <><, Pseudo Engineer, itinerant idiot