Re: several messages about Netscape and charsets

From: =?ISO-8859-2?Q?krem=BEsk=E1_HO=D8=C8ICE?= <Barry.Bouwsma@TUKE.SK>
Subject: Re: several messages about Netscape and charsets
Date: Fri, 15 Dec 1995 21:06:55 +0000

Next Article (by Date): Jeste jednou k lokalizaci Windows Ivan Janda
Previous Article (by Date): Re: Najde se nejaky named/DNS guru? Matus Uhlar
Top of Thread: Re: several messages about Netscape and charsets guest
Next in Thread: Re: several messages about Netscape and charsets The Radio Prague Staff of Highly Skilled Experts
Articles sorted by: [Date] [Subject]

This is to follow up on a message I posted last weekend...
 
> >  (But wouldn't these tags confuse
> > browsers, which use non-ISO-Latin2 fonts, like Netscape for MS-Windows?
 
    Having just had experience with this Netscape beta version, the
present beta of Netscape doesn't quite work right.  You could point your
finger at Micro$oft for making use of a code page which is different in
certain code positions from the ISO_8859-2 code page which is the
on-the-wire standard for Central European data.  The browser is not
confused (actually, it will try to display Latin-2 characters if your
default charset is otherwise, so it's working, to an extent).
 
    Netscape performs the switching based on the MIME charset tag, which
can be provided within a document by the META HTTP-EQUIV tag as mentioned
in previous messages and in the Netscape 2.0beta release notes.  What
Netscape recognizes to select the Latin-2 encoding is ISO-8859-2 (or some
unregistered Mac font), which will then display the page based on the font
you have specified under your preferences.
 
    In my test, with the Central European Windoze that I suspect sees
wide usage, the fonts which were available to me were a font with Latin-1
characters, which corresponds to the ISO-8859-1 MIME charset with no
problems, a CE font, and a Cyrillic font, and others I did not pay much
attention to.
 
    The problem is that Central European Windoze ships with a CE font
which doesn't quite match ISO 8859-2, and if people select that font,
certain characters are displayed incorrectly, as Slovak and Czech users
are well aware.  If one were to provide a font with the correct characters
to match the ISO 8859-2 definition, this would not be a problem.  But
that's not the way the system I was using was configured.  I'm not aware
of a font that matches ISO 8859-2 which one can use with Windows, but I'm
sure there is one available somewhere.
 
    The Windows code page for Central European languages is CP1250.  This
is mentioned in RFC1345, as well as a package of charsets I retrieved from
Germany.  Offhand, I do not know if the definition provided in RFC1345 is
entirely correct -- I do know that in the package called trans100, some of
the characters which appeared on the screen were different from the
description of these characters.  Furthermore, some widely-distributed
documents claim incorrectly that this code page matches the ISO 8859-2
standard.
 
    Because it is to be preferred that data delivered over-the-wire make
use of the ISO 8859-2 encoding standard, rather than a platform-specific
standard (or non-standard), my opinion is that Netscape should recognize
this encoding and be able to translate from this to CP1250 (as CP1250 is
not understood by Netscape as a MIME charset, even though it is
registered by virtue of being included in RFC1345).  I think CP1250
contains all the characters in ISO 8859-2, with some of them in different
positions, so a mapping should not be a problem.  I would rather see this
than see documents delivered with MIME charset tagging of CP1250.  As the
MIME RFCs note, as small a number of charsets as possible is preferred.
 
    I'm sending this to the Netscape for Windows bugs address as well as
posting to CSINFO-L/cz.net.csinfo, to offer this as a suggestion, which is
why I am explaining things that are well-known -- I do not know if Windows
fonts can be distinguished by encoding the way X fonts can, making it
possible to provide translation for CP1250 but leaving some
ISO-8859-2-compliant user-supplied font untranslated.  If not, then I
would suggest hardcoding the translation, since CP1250 is probably as
ubiquitous with Central European Windows as ISO-8859-2 should be as the
preferred Internet format.
 
    Again, the page previously noted with the META HTTP-EQUIV MIME
charset tagging of ISO-8859-2 found at  http://www.vszbr.cz/~guest/  will
display Latin-2 characters properly under X with ISO Latin-2 fonts, but
certain chracters are incorrect under Windows and code page CP1250
(z-caron is one of them).
 
 
Barry Bouwsma
visiting MZLU Brno, CZ
 
 
>     No more than delivering the document without the tagging.  Ideally,
> the client (Netscape) can perform the mapping from the over-the-wire
> encoding defined in the META MIME charset tag, to the display font which
> is in use locally.  But I don't know if the ISO-8859-2 -> CP1250 mapping
> is in place in MS-Win Netscape, likewise with Mac Netscape.

Go to listserv.cesnet.cz LWGate Home Page.