can edit an expression, and whenever it's parseable, it'll
update what it renders to. This necessitated making the core
memory-leak free even through tokenization and parsing errors.
I wrote some simple memory profiling hacks (just #define-d
aside malloc and free) to verify. Fun, fun. It even
handles (char *)-tagged malloc-ed blocks, postprocessed
usefully by (big surprise) a perl script. Subscript blobs
work too, now, (and both sub and sup parse -correctly-, as
opposed to not, which was the case yesterday) but
not subsup blobs at all. (ie you can't superscript
and subscript the same blob) Layout of sup and sub is still
extremely primitive. At 980 loc.
I can typeset the Axiom of Extensionality, whee.
Subsup blobs work, now, nominally. That is, I'm pretty
confident half of the parsing is correct, (a_b^c only)
but the layout is still shitty. At 1045 loc.
Need to get back to ptkfonted & ricercar hacking soon.