On 2010-11-15, Jürgen Spitzmüller wrote:
> Guenter Milde wrote:
>> Some comments:
>> * Does
>> # InternalEncoding <true|false>
>> relate to a font-encoding switch with Babel?
> yes.
>> Then it should be true also for e.g. languages with cyrillic script
>> (even if T2A - T2C are standard font encodings, the font-encoding
>> switch is hard-coded in Babel and interferes with XeTeX/LuaTeX).
> This bool is currently (only) used in
> Paragraph::Private::latexSpecialChar to avoid the use of latexSpecialT1
> when such a language is used within a T1- encoded document. It cures
> some bug, but I do not remember right now which one.
> It might be that we need to add some more languages, yes. But we will
> have to re-check first if it really is necessary.
So I suppose it might rather indicate a non-standard (i.e. non T*)
font encoding which has problems with "fake" symbols like
\DeclareTextCommandDefault{\textcopyright}{\textcircled{c}}
that expect an ASCII char to output itself.
The name of this key should reflect this (if my assumption proves true).
>> * Encoding <default_encoding> (not with XeTeX/LuaTeX)
>> is the default of the LaTeX source (for LaTeX the "input encoding").
>> The keyword name should reflect this (in our Unicode-times) rather
>> restricted meaning. (I prefer utf-8 as default LaTeX input encoding
>> if the users locale indicates utf-8 is in use on the system, but this
>> is yet another change.)
> I'm not sure I understand you here. The encoding is not the default, it is
> used whenever we switch to said language.
Only, if the LyX output encoding is "Default", which translates to
"use the language-default encodings for each language".
I argue that the use of multiple encodings in one file results in a
"unfriendly" LaTeX source. It does save some bytes but makes it very
hard to view/edit and impossible to re-encode the source.
This might be required for "pre-Unicode" CJK or Arabic, but is bad for
languages using Latin, Greek, and Cyrillic scripts.
> And for classic LaTeX, I thing we should stick with these encodings for
> now. I don't consider inputenc's utf8 support mature enough.
I would remove the iso-* and koi-* encoding settings.
inputencs utf8 support is rudimentary but actually quite mature. It even
avoids bugs in older encodings (like the conversion of the micro sign µ
to a mathematical \mu). Combined with LyX's strong Unicode->LaTeX support
it makes for a good choice on a UTF-8-based system.
>> * Language codes:
>> the "best current practice" for language codes on the Web advises to
>> use country codes only if necessary to disambiguate.
>> http://www.rfc-editor.org/rfc/bcp/bcp47.txt
>> http://www.w3.org/International/articles/language-tags/
>> I suggest to drop the country codes whenever the default variant of a
>> language is used.
> I'm not sure. We need these codes, amongst others, for the spellchecker and
> the thesaurus. Changing the codes right now could break working things.
Than this should be documented.
> As said, I do not want to change more than necessary at this point of the
> development cycle. Refinements can be done for LyX 2.1.
OK.
Günter
