Knit record cache considered harmful

[Michael Ellerman]

On 6/23/06, John Whitley <whitley at acm.org> wrote:
> Aaron Bentley wrote:
> > Many people have noted that building large trees eats gobs and gobs of
> > memory.  I suspected the problem was knit caching.
>
> Hmm, sounds like this might be the root cause of bug 43702 [1] for
> bzr branch throwing a MemoryError, where the working tree in question
> was about 1 GB.  I can certainly give this a run on my test case to
> see if turning off knit caches fixes this problem...

Yeah, I think that's probably just a dup of
https://launchpad.net/products/bzr/+bug/45310, except the former
reporter has less memory and so runs out, whereas John has plenty and
just eats up swap.

If you check the graph, we go from using ~1GB to ~100MB.

cheers