[Gnome-print] Re: [jody: [ynakai@redhat.com: Re: Gnumeric bug 15607]]

[Lauris Kaplinski <lauris helixcode com>]

Hello!

I consider it very bad idea, to accept current miserable locale/encoding
scheme at all.
Instead we should:
1) Use only utf-8 strings in application code
2) Collect all input/output related code to single place (like EFonts
are in gal), where it can be dealt in systematic way.

The current de-facto system of locale-specific encodings is way too
error-prone. To make is waterproof, one should include encoding tag
with every string passed in system - which results in much-much more
inconveniece than using utf-8 everywhere.
How can application know, what given character code really means? How
to determine, whether it is in iso-8859-1, koi-8 BIG-5 or whatever
encoding? You shouldn't force people using only 1-2 languages in
their documents - and what is even worse - you shouldn't expect that
received document is written in the same locale, where the reader
happens to be.

Locales/encoding are EVIL. They have to be replaced by languages/unicode
ASAP - and I think it is better to introduce major short-term pain,
that to make people suffer forever. The situation is bad enough, that
people browsers, mailreaders etc. have to support silly encodings
forever - better to not generate more such unhappy applications.

Lauris

> OK, I'll do all what I can do, because I have a full time job at RH Japan.
> 
> But remember, you can easily test the multibyte strings because there are
> also all English alphabets A, B, C in multibyte code. There is Japanese
> locales by default in glibc upper than 2.1.9x (glibc 2.2 beta) and Japanese
> fonts in the normal XFree86. 
> 
> Matt Wilson (msw@redhat.com) is very good at how English people can test
> apps in Japanese environment, because he had made Japanese Red Hat Linux
> before I joined to RH in the last May. He seems to be also one of the
> maintainers of cvs.gnome.org, so some of you may know him.
> 
> > > Do these patches make sense ?  Gnumeric is obviously not utf8 aware
> > > yet, so I'm unclear how changing the font selector fixes anything.
> > 
> > utf8 awareness is required for stepping in a string, otherwise any
> > application can deal with utf8 data, it just does not really know what
> > it is handling.
> > 
> > The fontset loader change seems simple and in line with what we have
> > done in the past for other gnome apps. 
> > 
> > But I am very confused about the style.c patch, a good description of
> > what is going on, and a detailed comment in the source code for the
> > maintainers would be the only way to go.  
> > 
> > The gnumeric renderer does not know how to handle multi-bytes yet
> > (last time I checked that code had been dropped, we tracked it down to
> > the printing merging code I think).
> 
> If you're sure some of you have some time to evaluate and accept
> the multi-byte patches, I'll rewrite them to the cvs version
> of gnumeric and send to you.
> 
> Thank you very much to give me a chance to what I'm really suffering...