handle memory leaks

[dsadinoff]

In regards to https://rt.cpan.org/Public/Bug/Display.html?id=93693
It's probably not that surprising that there's some real leaks in this module, expecially $my->{$self} = ...

attached is an implementation of a reset() method to clear the memory.

[mgreter]

I can confirm the leaks. I tried to add an automatic way instead of needing to call reset myself (since it would reset all instances). HTML::TreeBuilder uses Scalar::Util::weaken (if available) to solve circular dependencies with parent nodes. It actually turned out to be much easier. Just had to add a deconstructor to free the global resources stored in $my: mgreter@4f8fd2a

[dsadinoff]

Yes, that's a much better approach.

At least my leak test seems to hold up.

Cheers!

[mgreter]

I guess this can be closed as these commits were already included with #5.

[ingydotnet]

I think these changes got included in the application of #5.

Everything should be on CPAN pQuery-0.12

Thanks!

[mgreter]

@ingydotnet: I guess there was some confusion about this issue. I see you've added the documentation for the reset method back (310cc92). This is no longer needed as I was able to solve this by implementing DESTROY for the object, so the resources should be deallocated automatically! See my other commit from #5 (mgreter@3c57a37) which removed the reset method again! Basically what's left from @dsadinoff are the test cases. I guess other than that there was no real use for the reset method!?

It might be worth mentioning that this will only work (afaik) if weaken from Scalar::Util is available. pQuery could enforce that for HTML::TreeBuilder.

[ingydotnet]

@mgreter, can you possibly update the doc for me? It's in doc/pQuery.swim. Other docs get generated by it, but I'll do that part. The swim format should be fairly self-explanatory.

I didn't really go over the code, since I'm working on a ton of other stuff, but I want to get the contributions released.