Am Dienstag, 27. September 2016 um 20:02:26, schrieb Guenter Milde <mi...@users.sf.net> > Dear Kornel, > > the current tests do not include non-default input encoding. > > IMV, testing export with inputencodings "utf8" and "ascii" would help to > find several hidden bugs and make LyX more robust. > > * While the inputencoding is not important as long as the *.tex file is only > temporary, "utf8" and "ascii" have use cases > > * when exporting to LaTeX (collaboration, archivation) > * for debugging > > LyX's default "mixed language specific input encodings" is especially > unsuitable in these situations. > > * Some Babel-languages expect "utf8" as default, > e.g. Babel-Russian recommends utf8 since several years. > LyX still uses koi8-r because switching to "utf8" without > previous tests is too riscy.
I tried to use UTF8 for ru/Tutorial.lyx. pdf4_systemF seems to work with 'Dejavu' fonts. pdf4_texF does not work. That is, I don't know which tex-font should I select to make it working. > * Testing "ascii" would allow us to separate problems due to the XeTeX > engine from encoding problems for failures of ".*pdf4_texF". > > * With stable and tested export to utf8-encoded LaTeX files, this could also > become a default for Latin-script based languages. > > > I expect the tests to uncover some problems similar to what we have seen > with XeTeX-texF. > > There is no need to test this with all export formats, it would suffice to > test with pdflatex (pf2), say. > > > What do you think? If we can specify the encoding in the lyx-file depending on language and output format, then it is easy. In some cases we already would get encodings from the language file 'lib/languages'. (ATM disabled) See export.cmake:34 and useSystemFonts.pl:146. I am open to suggestions. > Günter > Kornel
signature.asc
Description: This is a digitally signed message part.