[pygccxml-development] parse time with large projects

[Kevin W. <kev...@gm...>]

I'm using "cache = directory_cache_t('cache')" in my module_builder_t
constructor, and it does cut down on the parse time that would
otherwise occur, but it still takes around 10 seconds for each .h
file--and since my project is only growing in size, the time it takes
to get through all the header files is becoming prohibitively long.

Before I go hacking on the cache stuff, I wanted to ask here--is there
a specific reason why the hash of a header file, plus all the
preprocessor defines and include paths, are not enough to recognize an
already parsed header file? Recognizing these should be an operation
that takes less than a second, I would think--but some of them take as
long as ten.

Does anyone else have any suggestions on cutting down the turnaround
time for a growing Py++ project?