[opensource-dev] ok wtf is this?

[Tateru Nino]

On 5/09/2011 4:47 PM, Dave Booth wrote:
> On 9/5/2011 12:53 AM, Tateru Nino wrote:
>
>> In order of likelyhood (most to least), I'd say: corrupted texture
>> header data, corrupted heap, fragmented heap.
>>
> After a few moments of reflection of my own, I agree with your 
> assessment..  The question that springs to mind is why does it happen in 
> an area where I know all the textures have been loaded successfully in 
> previous builds without any of the three issues showing up?
All bets are off with a heap-corruption bug. You might have clicked
twice, in rapid succession, on a UI widget with a non-reentrant handler,
or you might have viewed some _other_ content at some prior time during
the session which caused an allocation overrun. Whatever causes the heap
to mess up lays around like a time-bomb waiting for the allocator's pool
manager to discover that things are "messed the hell up".

Heap-corruption bugs are a pest, since there's usually little or no
relation between where the app fails and where the bug _really_ is. It
might even be easily reproducible - but that doesn't necessarily narrow
it down any, alas.