[tex-live] packages with characters > 127
zdenek.wagner at gmail.com
Thu Dec 31 02:29:00 CET 2009
Imagine that someone would write something in French using UTF-8 and
load a .sty file with ISO-8859-15 encoded characters in it. It won't
work. If the characters were entered as TeX sequences, it would work
in any case.
2009/12/31 Manuel Pégourié-Gonnard <mpg at elzevir.fr>:
> Karl Berry a écrit :
>> It is done (not the TL09 version, though). The point is, packages using
>> non-ascii characters currently don't take that kind of precautions.
>> The point for me is that it shouldn't be up to each individual package
>> author. It should be done at the engine level.
> If it were doable, I would perhaps agree, but I hardly see a way. (Except
> creating a "bytes mode" in XeTeX/LuaTeX and making it the default, which would
> IMO be a greater evil that the small problems we're facing now.)
>> The point for me is that the reality is that authors don't stick to
>> plain ASCII and there is no real reason they should. If a French author
>> wants to use 8859-15, that should be ok. We should be liberal in what
>> we accept.
> Already with current latex/pdftex, they can't without restrictions. The
> restrictions will only be slightly higher with LuaTeX/XeTeX.
> (If "we" is understood as "TL/ctan2tds", then I agree that we should be liberal
> and treat everything as bytes in an encoding-agnostic way and without altering it.)
>> The point for me is that, again, with such a basic thing as character
>> encoding, there should be no issue of "compatibility" at all. Authors
>> have preferred character sets. We (implementors) should deal with that,
>> not to try ram the One True Way down everyone's throats.
> I don't see asking people to declare the encoding they use (or at least whether
> bytes mode or some-variant-of-unicode mode is to be used) as an unacceptable
> pressure on authors.
>> Like line endings ...
> Line endings are technically easier :-)
More information about the tex-live