[Pmccabe] new set of fixes for pmccabe

[Matt]

On Thu, 12 Aug 2010, Alfonso Cepeda wrote:

> pmccabe 2.6, (now in debian unstable) has included your patch.
>
> Thanks for your help.

I have created a new bug and uploaded a 'patch':
https://bugs.launchpad.net/ubuntu/+source/pmccabe/+bug/659550

I fixed several parsing issues, some of which were mentioned by the 
pmccabe mailing list:
1. recognize throw(), volatile, compiler attribute, and other 
function/method declaration keywords.

2. skip code in #else/#endif blocks

3. skip code in #if 0/false blocks

4. fix various hangs/incorrect parsing in really weird, but still 
legitimate code (see added tests for details)

5. fix a few minor memory corruption issues (found by valgrind), some 
latent and some introduced

6. sped up by 80% by using mega-compilation and whole-program optimization

7. improve diagnostic messages to reduce debugging time:
"got EOF"
versus the improved
"expected ';' before EOF"
and
"expected ')' before end of line"


The primary issue that remains (and is documented in the TODO), but 
doesn't adversely affect the results on the codebases I tested, is the 
scenario where a #else block is for a #if 0 block, both blocks will be 
skipped. Someone else can take up the mantle for that one :)

I added new tests to the testsuite for everything I fixed, and 
successfully ran pmccabe against the sources for gcc, scummvm, mockitopp, 
amop, pmccabe, and private projects with no hangs or valgrind-detectable 
issues. I also manually verified that the top and bottom 10 functions by 
complexity were reported accurately.

Please let me know what I need to do to get this commited to CVS and/or 
integrated into the Debian/Ubuntu package.


--
tangled strands of DNA explain the way that I behave.
http://www.clock.org/~matt