On 2015-12-13, Kornel Benko wrote: > [-- Type: text/plain, Encoding: quoted-printable --]
> Am Sonntag, 13. Dezember 2015 um 22:25:34, schrieb Guenter Milde > <mi...@users.sf.net> >> On 2015-12-13, Kornel Benko wrote: >> > Am Samstag, 12. Dezember 2015 um 22:52:13, schrieb Guenter Milde >> > <mi...@users.sf.net> > ... >> >> c) consider excluding more output formats based on BufferParams, e.g. >> >> * pdf[245] and dvi3 if the language is Japanese >> > Such things are difficult to handle. It suggest the same logic be >> > implemented in tests and in BufferParams. >> I was thinking of LyX behaviour. It seems this is already implemented. >> Then, the test should not recreate the logic but test it: assert the >> excluded formats fail with a sensible message. > Yes, but to know which should be excluded is the problem for tests. Of course, but IMO it is important to have it independent in the test -- to verify the built-in logic. >> >> * XeTeX with TeX fonts unless the inputenc is ASCII >> > We are setting it already in that case. >> I only implemented the forced switch to ASCII because I did not manage to >> exclude XeTeX with TeX fonts for inputenc != ASCII >> (the behaviour I prefer over the current way - overriding the user >> selection). > I meant the test suite is setting ascii for xetex + tex font. This is superfluous since LyX does it, too (since my fix to #9740). So this fix was not tested due to "massaging" in the test machinery and there may still be cases where it fails :-( Günter
