On 2019-02-06, Scott Kostyshak wrote: > On Tue, Feb 05, 2019 at 09:18:16PM +0100, Kornel Benko wrote:
>> > > ... >> > > After this commit, test with TL18: >> > > $ ctest -R unicodesymbols >> > > --> (960 tests) >> > > The following tests FAILED: >> > > 442 - >> > > INVERTED.DEFAULTOUTPUT.TODO_export/export/latex/unicodesymbols/001-4-latin_cp862_pdf2 >> > > (Failed) >> > > 448 - >> > > INVERTED.DEFAULTOUTPUT.TODO_export/export/latex/unicodesymbols/001-4-latin_cp1255_pdf2 >> > > (Failed) >> > > 449 - >> > > INVERTED.DEFAULTOUTPUT.TODO_export/export/latex/unicodesymbols/001-4-latin_cp1256_pdf2 >> > > (Failed) >> > > 458 - >> > > INVERTED.DEFAULTOUTPUT.TODO_export/export/latex/unicodesymbols/001-4-latin_iso8859-6_pdf2 >> > > (Failed) >> > >> > Could you check, whether the output is correct (i.e. the same as with >> > inputenc == utf8 or inputenc == ascii) or whether there are wrong >> > characters? >> > >> > If the output is OK, this is fixed by preventing the fallback loading of >> > utf8 and we can remove the "inversion-clauses". >> Totally wrong characters displayed for latin-1 supplement and also for latin >> extended. >> (All of the created pdf''s are unique) Thank you Kornel for the ("private") PDFs. I fixed lib/unicodesymbols, could you re-check? > These tests aren't failing for me. Maybe you did not re-compile since 32e4a85e4 (3 days ago)? Anyway, I fixed the "unicodesymbols", so these tests should now compile (i.e. the inverted tests fail) even without re-compiling. If not, I have missed some force flags. > Note that for me the following test is failing: > ctest -R > "DEFAULTOUTPUT_export/export/latex/systemfonts-Math-missingchars_pdf5_texF" Yes. This test is not intended to run with 8-bit fonts. (And I am convinced that any documentation and test that has explicitely set non-TeX fonts is expected to fail (or pass by pure accidence) with TeX fonts. Kornel, could you change the test scripts to not replace non-TeX fonts with TeX-fonts, please? Günter