Main change is support for even more TeX.
= More TeX =
Of 6302 tests: * 5497 (87.2%) passed * in 5072 (92.3%) cases it was able to produce HTML * in 425 (7.7%) cases it couldn't, and I doubt this could be significantly improved. * 2 (0.0%) failed due to lexer (and in both cases they should fail) * 20 (0.3%) failed due to parser, in all cases due to abuse of ^s and _s. Could be fixed, but it's not a priority. * 783 (12.4%) failed due to 203 unknown -codes * 179 -codes in 684 (10.8%) equations were illegal (couldn't be compiled by LaTeX with ams* packages) * 24 -codes in 99 (1.5%) equations were legal * So 121 (1.9%) failed due to reasons other than genuinely illegal -codes * So of 5618 equations not containing genuinely illegal -codes, 5497 could be compiled by texvc, what makes up for 97.8%, close to promised 99%.
= Performance =
6302 equations, 119 145 bytes total, are all parsed in:
real 0m0.191s user 0m0.180s sys 0m0.000s
real 0m0.190s user 0m0.190s sys 0m0.000s
real 0m0.196s user 0m0.190s sys 0m0.000s
on Athlon 1 GHz box. These are results of texvc_test, real results will be slightly lower due to overhead of starting texvc on every equation.
texvc will introduce no performance problems, and in fact can be faster than "just run TeX" solution as it can find out which TeX modules should be loaded.
= Open issues = * transparent PNGs * Unicode support * checking which symbols need AMS to be loaded