Devel::LeakTrace

[Richard Clamp]

Andy Armstrong wrote:
> I started looking at Devel::LeakTrace[1] a couple of days ago. It 
> wouldn't easily build on my Mac because it needs glib (it uses glib 
> hashes internally) so I chucked it on a Linux box and - while it seems 
> to do what is says on the tin - it makes any non-trivial script run like 
> a dog.
> 
> So I decided to see if I could fix it. I replaced its dependency on the 
> glib hash with a standalone hash implementation I'd written for another 
> project, got it working correctly (I think - the tests are broken 
> too...) and started investigating the performance problems. To cut a 
> long story short the currently implementation is roughly O(N^2) with all 
> that implies...
> 
> Anyway, now I'm this far in I might as well fix it properly - but that's 
> basically going to mean a complete re-write - to be blunt it's not a 
> large module and most of it is currently wrong :)
> 
> I mailed the author, Richard Clamp, a couple of days ago but have had no 
> reply. If I can't get hold of him by the time I have a release ready 
> what should I do?

Continue to wait.  I got your mail yesterday and haven't even read it 
much beyond flagging at as "get to this week".  Nice to know that what 
I've done is completely wrong though.


  I could release it as a different module - but that
> seems to be a needless pollution of module namespace. What's the 
> protocol for replacing a module which - although it sort of works - is 
> unusable in most practical situations?
> 
> [1] http://search.cpan.org/~rclamp/Devel-LeakTrace-0.05
> 
> --Andy Armstrong, hexten.net
> 


-- 
Richard Clamp <richardc at unixbeard.net>