> You will need to build samples and tests and then run sanityTest.pl
> located in scripts/ on the data found in samples/data. I will also
> run our internal tests to make sure there is no performance
> degradation.

Got it.

> Oh, it is actually more complicated than I thought. I don't think it
> will be wise to back-port all this DOMMemoryManager-related changes.

I wasn't sure either.

> How about a simplified version, that boils down to the following:
> 
> 1. Split kHeadAllocSize into kInitialHeapAllocSize and kMaxHeapAllocSize
>    with values from 3.0 branch.
> 
> 2. Implement exponential growth until kMaxHeapAllocSize as in 3.0 (for
>    that you will need to add the fHeapAllocSize variable to the
>    DOMDocumentImpl class).
> 
> While there won't be a way to tune fHeapAllocSize, the initial small
> value and the subsequent growth should satisfy most applications. What
> do you think?

I think that's fine, but as I said, I'm the one who needs the lower values.
;-)

It will take three allocations to get the value back to 64k where it is now,
so I don't imagine that will hurt most applications too much, particularly
if your other DOM patch really speeds up large document parsing by 20%.

I'll prepare a patch against the 2.7 branch with your suggestion and test
it.

-- Scott



---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to