Re: Updated HOWTO

[Jean-Marc Desperrier <jean-marc.desperrier@xxxxxxxxxxxx>]

Erik van der Poel wrote:

> Bruno Haible wrote:
> >
> >   - Note that the Cyberbit font is not free.
>
> Bitstream Cyberbit is available:
>
>   http://home.netscape.com/eng/intl/basics.html#fonts
>   ftp://ftp.netscape.com/pub/communicator/extras/fonts/windows/

But not "free".

Microsoft's Arial unicode is also "available" and more complete :
http://officeupdate.microsoft.com/2000/downloadDetails/aruniupd.htm

> > - Browsers:
> > - Mailers:
>
> Some support for UTF-8 in Netscape 4.X, and more support in Netscape 6.

Erik, you know very well the UTF-8 support of Netscape 4.x has many, many
bugs.

It seems it has significantly less bugs in the browser under Linux than it
does on other platforms, but it has at least the invalid UTF-8 character
sequence generation bug in mail .

A description of this bug for those interested :
When creating a mail document in Netscape 4.x and sending it in utf-8
format, in sequence of utf-8 characters that ends with a value 0xA0, the
final 0xA0 is replaced by a character 0x20.
When this 0x20 occurs at the end of a line, it is supressed.

This results in illegal utf-8 sequences :
 For example, the french character "à" that should be encoded as 0xC3 0xA0
is encoded in messages sent by Netscape Messenger 4.x as 0xC3 0x20.
This also happens with another greek character, and I've seen it also with
some japanese characters.

I've reported the consequences of this bug on Mozilla in bug 35004, and
Mozilla now handles this invalid sequences gracefully, but I've never had
an answer from Netscape about correcting the original bug in Netscape 4.x,
and for performances reasons, most people will not update to
Mozilla/Netscape 6 as of now.

> > - Editors:
>
> Again, Netscape 4.X and 6.

None of them is an _editor_.

> > I'd invite all of you to send me (or the list) small notes about new
> > package releases that work in UTF-8 locales or (even better) in all
> > multibyte locales.
>
> Netscape 4.X and 6 support a number of locales, and can be extended to
> support some more by editing a text file.

Interesting, but I've never seen this documented until now.

-
Linux-UTF8:   i18n of Linux on all levels
Archive:      http://mail.nl.linux.org/lists/