[texhax] Any crazy math formulas for testing a TeX language interpreter

[Douglas McKenna]

David Carlisle wrote -

> Actually for the base latex tests that should work well as it is designed to run with a sandboxed tex input path (directory) _not_ pulling in the local tex tree, so we know what we are testing. 

Makes sense.  Good to know.

> luatex's aren't much teh same either, but even if they are completely different, like any regression test suite you only need to sign off the base results once and engine-specific logs can be stored, then it will flag if something changes. 

Yup.  The log file format has to be stable enough to be compared with earlier versions of itself.

> PS I feel I have to ask if you could change the name, which would be _very_ confusing: anyone who sees that name is going to assume it is related to JavaScript and the css box model, but I gather it's related to neither. (I would have assumed that even if it were not the case that there is an existing jsbox project on github that is about that)

Understood, and that has been the plan (some kind of name change).  JSBox was just a goofy joke, since I enjoy playing a certain composer on my piano.

> Do you just need examples using classic TeX fonts or does the system also accept OpenType Math fonts using the MATH table (as xetex and luatex do)?

Right now, just TFM and just Computer/Latin Modern.  I'm not finished with the OpenType parser yet, although I am with the part dealing with the 'MATH' table.

> If you can get the unicode-math package to work that would be a good test of many of the features of the OpenType math fonts (such as latin modern math)

Agreed.  But I'm trying to keep the complexity of the font world at bay until it becomes unavoidable.  One step at a time. :-)


- Doug McKenna