Just on plugins and Mata and such, Mata also has one major advantage
over C. Compiling via Stata means that the source code can be
practically distributed rather that the binary code, which means not
only cross-platform stability, but security through transparency (not
that everybody will read the code, but just because they can).
I've tossed up the idea of posting on statalist to query about
volunteers to compile C code on different platforms, but decided not to
for security reasons. The only binary code I trust myself to distribute
to others is the code I compile myself. Even on the friendly statalist
there may be the occasional subversive.
If StataCorp were feeling like they have (vastly) too much spare time,
it would be cool to make some sort of a cleaner arrangement with
compiling plugins. Through just skimming through the Mata manual I see
particular areas where it seems they've taken the major C libraries and
cut bits of them out, but oh how those lost functions can be useful.
Indeed if a better C arrangement were available, Stata could be extended
in such ways as, for example, allowing bits of useful GIS functionality,
or making available Stata API's for different types of database server.
I think people stay away from it because plugins in Stata are scary and
platform dependent; a pity, because they can do things that Mata will never.
What would be perfect is if every Stata release came with C and FORTRAN
compilers that are perfectly cross-platform... though of course reality
prevents that.
Anyway, the more I think about Mata the more I like it.
