[Aeditor-talk] regexp, yet another redesign

[Simon Strandgaard]

Want to know about current status of AEditor, then read on.

During January I have made a more correct working engine (yet not
working fully correct). There is still lots of problems with nested
quantifiers.

For instance /(a.*b){3}/ has both 
an outer expression  ( ... ){3}
and an inner expression a.*b

Such kind of expressions requires a smart data structure.

Earlier my problems were that the data structures was too simple.
My current problem is that its too complex, and I cannot conceptually
overview it (I wish an upgrade of my puny brain for my birthday).
The current implementation (see 'repeat_nested' branch in CVS) cuts
pieces off the timeline, and reinstall them later. This removal/install
operations on the timeline doesn't fit with the metafor I know from
nature. I need something more natual, which I can cope with and easy
understand.

Ok, something in between is needed. Last night, while drawing and
analysing, I got a new refined idea for a new data structure, which I
hope will do better.

At least I have become better at unittesting and upgraded my debugging
skills with +2.7 levels.


When nested quantifiers is working (correct), then I will finish the
implementation of perl6 regexp, where Mark Sparshatt already have done
some work. Don't know what condition the perl6 implementation is in
though.

Then we kick ass... but its some units more out on the timeline.

--
Simon Strandgaard