Axel Thimm | 13 May 2003 16:45
Picon
Picon
Favicon

Use always -translate-file=cp8bit?

On Tue, May 13, 2003 at 03:22:27PM +0200, David Kastrup wrote:
> Axel Thimm <Axel.Thimm <at> physik.fu-berlin.de> writes:
> > No, I didn't override, but you seem to be right about the problem with
> > 8-bit cleaness. The short version of the configure test:
> > 
> > latex '\nonstopmode\message{^^fc!}' 2>&1 | grep 'fc!$' > /dev/null  && echo BAD || echo GOOD
> > 
> > does fail in my locale (en_US.UTF-8, default on Red Hat since Oct
> > 2002). Prepending the test with "LC_ALL=en_US.ISO8859-1" passes the
> > test.
> > 
> > So utf8 locale is not supported?
> 
> By LaTeX.

> The manual entry "LaTeX international characters" will tell you how
> to go about this stuff.  As long as LaTeX is not comfortable with
> UTF8, I'd recommend to use Latin1 as your default language
> environment, anyway, but you can also make do by using the
> -translation-file option as described in that section.

Checking the matter further, I don't think that anything but
-translation-file is a good solution to that problem.

I think I am starting to understand preview-latex internals: The TeX
run output needs to be in the same encoding like the TeX input for
matching the previews.

Now consider mixed inputencodings in the same file (e.g. mixed
latin1/greek/arabic). The TeX output is controlled by the locale which
is a global parameter concerning a single run. So the switching of
inputencodings will not be respected, unless one uses the cp8bit
pass-through method.

Even if/when TeX itself becomes utf8 aware concerning the TeX run
output, and even if the ucs/utf8 encodings should enter the LaTeX
kernel, you will still have the problem of matching mixed input
encodings.

Doesn't that mean that preview-latex should always use
-translate-file=cp8bit?
--

-- 
Axel.Thimm <at> physik.fu-berlin.de

Gmane