Jean-Marc Lasgouttes wrote:
> Question: is the equivalent of \'{e} '+e or directly é (in the
> internal representation)? I do not understand the stuff about
> surrogate characters in the patch.
lyx2lyx will convert \'{e} to é. But since '+e (or rather e+' - the accent
comes after the accented char in unicode) is also possible in ucs4 encoding
we have to support both. The latter could for example come from external
sources (clipboard, plain text import) or the unicode insert lfun. The
alternative would be convert all e+' that come from external sources to é,
but that would be more messy, and qt4 does handle e+' just fine.
> Second point: rather than putting in explicit preamble snippets I
> think we should go for feature (i.e. put "textcomp" there, and handle
> it in LaTeXFeatures).
That would be faster, but it would not be possible anymore to use arbitrary
packages. We will never be able to ship a complete list (if you look e.g.
at plastex it lacks many accents, although it has really a lot of symbols).
Therefore I think it would be a good if users can add symbols even if they
come from a specialist package (e.g. for ancient african languages or
something like that) we don't even know.
Maybe we can do both: use a feature for known packages, but allow to load
other packages as well.
> Is it really necessary to look at the outerfont to know the encoding?
I hope that it is not, because outerfont is probably one reason for the
slowdown.
> Can environment nesting really change the encoding?
Can it change the language? If yes, it can also change the encoding. If not,
then we can leave it out.
> Another question: what is the status of kmap files with this patch?
Not changed. I don't know the kmap related code very well, but I did not
touch that code at all. If it worked before it should still work.
Georg
