Pedro Andres Aranda Gutierrez <paag...@gmail.com> writes: > attached is an updated version of the multilingual demos that takes > the evolution of the code into account. > I'd really appreciate more people feel like giving it a look/try and > providing feedback.
Hi, Tested multi-polyglossia.org again on the updated feature branch. Looked good! The Arabic is now properly cursive. (The commas are the wrong way round, but I think that's from the ipsem-lorum source text used, and has nothing to do with your code.) However, it turns out it's /not/ actually working as it should for Arabic and Greek: the specified fonts are not used. This bug is masked by the use of FreeSerif as the main font in the test document, since it covers both scripts. If you change the main font in .dir-locals.el to a Roman font that does not include these scripts, the Arabic and Greek sections fail with boxes instead of glyphs. Xelatex complains: ! Package polyglossia Error: The current latin roman font does not contain the "Arabic" script! (polyglossia) Please define \arabicfont with \newfontfamily command. And ditto for Greek, but not for Devanagari. So the bug appears to be that the \newfontfamily commands in the generated latex define =\arfont= and =\elfont= straight from the two-letter codes, not the expected =\arabicfont= and =\greekfont=. Changing these names in the exported .tex solves the problem. The other thing I see is that xelatex (but oddly not lualatex) complains a lot that packages in org-latex-default-packages-alist are loaded after bidi.sty (which is part of texlive-lang-arabic and a prerequisite). Doesn't seem to cause any gross errors, but there may be subtle ones (we'll need an Arabic-speaker to look it over eventually). Yours, Christian
