Rx had some time ago (~0.0.6) a state structure rxinfo. AFAIK this was
tossed for speed reasons, state is kept in registers now.
This has several impacts:
- regexen are not reentrant anymore (global intstack)
- they don't/can't manage their allocated resources like bitmaps
1) I would propose, that all ops, which work with bitmaps and rxstack
have again a state structure.
(Reentrante regexen would probably need all the state there, but these
could be separate ops as well)
2) For bitmaps I would provide a bitlist.c with functions for setting
and testing bits. This bitlist would be based on list, so it should be
fast enough and had no limits WRT unicode chars.
3) There are some global bitmaps like bitmaps of RX_WORDCHARS or
RX_SPACECHARS containing all word or space chars respectively. These
should probably go into the chartypes/ directory, currently they are
usacii only.
Comments welcome,
leo
- RE: rx.ops, bitmaps and memory leaks Leopold Toetsch
- RE: rx.ops, bitmaps and memory leaks Brent Dax
- Re: rx.ops, bitmaps and memory leaks Leopold Toetsch