[Pyrex] Pyrex 0.9: Compiled code produces segmentation fault

[Paul Prescod]

> -----Original Message-----
> From: Greg Ewing [mailto:greg at cosc.canterbury.ac.nz]
> Sent: Tuesday, January 27, 2004 3:16 PM
> To: pyrex at lists0.copyleft.no
> Subject: Re: [Pyrex] Pyrex 0.9: Compiled code produces segmentation
> fault
> 
> This might not be what's really happening -- it might just
> be a symptom of the heap having been corrupted by something
> else.

I agree. But in this case the memory manager is reliably reporting the error
on a particular line of C code. So I'm inclined to look at that line of code
first. ;) I haven't built a debugging Python 2.2 so I haven't investigated
it to the level that I'm totally confident it isn't heap corruption.

> ...
>
> I can't see anything wrong with your code, so it looks like
> it may be a Pyrex bug.
> 
> There was a problem reported recently concerning trouble that
> can arise if a garbage collection occurs during a __new__
> method -- this might be a manifestation of that.

As you pointed out, nothing is for sure, but the previous bug could be
reliably reproduced by putting a gc.collect() in the middle of __new__ and
this one cannot. Also the previous one was caused by calling around
constructors in an inheritance structure but this one has no inheritance.

 Paul Prescod