[Bioperl-l] A better fix for Species.pm memory leak, please REVIEW.

Sendu Bala bix at sendu.me.uk
Sun Nov 23 07:19:57 EST 2008

Chris Fields wrote:
> On Nov 22, 2008, at 6:21 PM, George Hartzell wrote:
> I'm not quite sure why we are attaching a Bio::Tree::Tree to the 
> Bio::Species, so I need to go back and review the changes from the 1.5.2 
> release (I'm sure there is a good reason, just can't recall and haven't 
> had time to look).

The point of the refactor was that a Species object no longer be a dumb 
(list of) scalar, but actually 'know' what its nodes are, so you can do 
things like compare Species and find the LCA and such. You need a 
Bio::Tree::Tree for that.

> However, we aren't planning on retaining Bio::Species too much longer 
> anyway (we'll be moving towards a simpler Bio::Taxon-based system after 
> the 1.6 release and will be deprecating Bio::Species eventually).  Any 
> fixes should probably take this into consideration.
>> I'll commit it if someone'll review it.
> I vote to go ahead and commit this (the solution seems sound), but Sendu 
> is the one who would be best to vet as he refactored this.  If there 
> isn't word by the middle of the week you can go ahead and make the 
> commit.

I doubt I'll have time to really look at it properly in that 
time-frame... my initial reaction is that it is a bit bizarre (we are a 
species object intended to hold information about the species that we 
represent, and in order to do that... we create and store another 
species object in ourselves?!)... but if it works, it works? Is there a 
test in Species.t that does a Species object comparison and finds the 
LCA? Does it trigger the problem, and does it still get the correct LCA 
after this patch?

But anyway, thanks George, go ahead and commit if it all seems functional.

If you could summarise your solution in the bug report(s?) as well, that 
would be great.

More information about the Bioperl-l mailing list