diff options
Diffstat (limited to 'to.do/README')
-rw-r--r-- | to.do/README | 46 |
1 files changed, 46 insertions, 0 deletions
diff --git a/to.do/README b/to.do/README new file mode 100644 index 0000000..7123c30 --- /dev/null +++ b/to.do/README @@ -0,0 +1,46 @@ +The contents of this directory are: + + Wilhelms.todo + This guy Gerhard Wilhelms did an exhaustive line-by-line + study of flex, finding a large number of glitches; it's + not clear how significant they are (some have subseuqently + been fixed). + + Wish-List + A long list of cryptic one-line descriptions of various + features people have asked for, or I've thought would be + handy. If you have questions about particular ones, let + me know. + + reentrant + A set of mods contributed by John Bossom for adding + an option to flex to make reentrant scanners. + + rflex-2.5.4-diffs + A set of mods contributed by Chris Appleton for + the same purpose. + + streams + email from David Madden about coping with non-blocking + I/O. + + unicode + patches to support Unicode scanners, contributed + by James A. Lauth. + + +Of these, the ones that people frequently ask about are support for reentrant +scanners and support for Unicode. I haven't played with the reentrant stuff +so don't know how solid / well designed it is. I've sent out the Unicode +stuff to numerous people and haven't received any complaints, so I imagine +it works well. + +Another thing people ask for fairly often is removing the limit on size +of the scanners. There's some mail in the faqs/ directory about ways to +do this by cranking some #define's in flexdef.h, though of course the best +solution would be dynamic memory/resizing. + +By the way, I have translated the core flex algorithms into a C++ regular +expression class (but one that doesn't support the uglier stuff like +yymore()/yyless(), trailing context, etc.). If you ever wind up wanting +it, just let me know. |