From: Eric M. Ludlam <eric@si...> - 2008-06-09 22:57:02
I have a few changes in CEDET/CVS to post about beyond some of the
bug fixes I replied to.
First, after working out some of Ming-Wei's issues, I chased down a
last inaccuracy in the handling of subclasses for the type of
sub-classing (ie, public, private, etc). This means an update in the
parser to remember the information, plus to the analyzer. I'd be
particularly appreciative if someone more familiar with the nuances of
C++ could check semantic/tests/testsubclass.cpp for testpoints 9
through 14 to make sure I have the correct interpretation of the C++
info I found online. Thanks.
I also update the semantic work function to parse neighboring files
to the buffer you are working on, as this seemed to be a requested
feature. Profiling showed horrid performance, so I added more
statistics into the database tables for file size and lastmod time (as
per conversation here some months back.) Once your files are parsed,
it is now quite fast. Be aware, however, that old save files will not
have this information in them. It should automatically update them,
but your mileage may vary.
Be on the lookout for ways to optimize the new work function and let
me know. It should be easy to start tuning the current simple
algorithm for ways to guess the best set of files that need parsing.
I updated the way SRecode apps add themselves to the SRecode keymap
and menu, and improved the get/set inserter for private class members
(C++). I'm not sure how many folks are using SRecode, but if you do,
this should be more usable now.
Eric Ludlam: eric@...
Siege: http://www.siege-engine.com Emacs: http://cedet.sourceforge.net