Am Dienstag, den 21.02.2017, 09:12 +0100 schrieb Jürgen Spitzmüller: > > Sounds good to me. > > I'll have a go at that.
Could you try the attached patch? This should be tested against all test files, since it omits loading of xunicode if tipa is not required. Thanks Jürgen
diff --git a/src/BufferParams.cpp b/src/BufferParams.cpp
index a4c3a80dc1..a725068792 100644
--- a/src/BufferParams.cpp
+++ b/src/BufferParams.cpp
@@ -2176,12 +2176,19 @@ bool BufferParams::writeLaTeX(otexstream & os, LaTeXFeatures & features,
os << "}\n";
}
- // xunicode needs to be loaded at least after amsmath, amssymb,
+ // xunicode only needs to be loaded if tipa is used
+ // (the rest is obsoleted by the new TU encoding).
+ // It needs to be loaded at least after amsmath, amssymb,
// esint and the other packages that provide special glyphs
- // The package only supports XeTeX currently.
- if (features.runparams().flavor == OutputParams::XETEX
- && useNonTeXFonts)
+ if (features.mustProvide("tipa") && useNonTeXFonts) {
+ // The package officially only supports XeTeX, but also works
+ // with LuaTeX. Thus we work around its XeTeX test.
+ if (features.runparams().flavor != OutputParams::XETEX) {
+ os << "% Pretend to xunicode that we are XeTeX\n"
+ << "\\def\\XeTeXpicfile{}\n";
+ }
os << "\\usepackage{xunicode}\n";
+ }
// Polyglossia must be loaded last ...
if (use_polyglossia) {
signature.asc
Description: This is a digitally signed message part
