On 2016-09-28, Scott Kostyshak wrote: > Using system fonts + LuaTeX with fr/linguistics.lyx, we produce the > following .tex code:
> C'est-à-dire, pour \textipa{\begin{english}% > [d̥] > \end{english}% > } vous devez saisir \texttt{<d>} et ensuite le sous-diacritique en > Can someone confirm that LyX is doing the right thing here? From what I > understand, the exported code is correct given that the character inside > \textipa is indeed marked as English in the .lyx file. There are several issues here: * Yes, if the character inside a inset is marked as English, it should be exported this way. * I consider it a bug (or enhancement request) that LyX - does not change the language of inset content together with the surrounding text - defaults to English for some insets (e.g. mathed) (Copy a part of the mathed content to a document in a language other than English). Suggestion: make the default langugage for insets "inherit", i.e. "use the language of the containing paragraph" (name and idea from CSS. > This error goes away if we do not use English for that character. Maybe we should generally disable language changes in a tipa inset (i.e. make "inherit" mandatory). > Note that there is still a missing character error but that (I believe) > is a separate issue. Yes. This goes away if you use Free* fonts. > If someone confirms that the above exported code is indeed correct (i.e. > it is the responsibility of the user not to mark a character inside an > IPA inset as a different language), then > I will remove the language switch for that character. I propose to fix fr/linguistics and create a minimal example with language-switch in a tipa inset in autotests/export/latex. Günter