Re: Memory leak due to bidi?

[Andy Moreton]

On Wed 03 Aug 2011, Stefan Monnier wrote:

>>> I've given up on using VC annotate in emacs after 10 minutes of bzr
>>> thrashing my machine with zero output.
>
>> ??? "time bzr annotate src/character.h" shows this for me:
>
>>   real    0m21.802s
>>   user    0m20.430s
>>   sys     0m0.650s
>
>> That's not long enough to be annoying, let alone 10 minutes of
>> thrashing.  What am I missing?
>
> Thrashing slows the process tremendously, so the 10 minutes are probably
> due to thrashing.  And thrashing is most likely due to Bzr using more
> memory than Andy has on his machine.

Sorry for being imprecise. I didn't mean thrashing in the sense of
spending all of its time paging in and out of swap. I meant that I had
one CPU pegged at 100% and emacs unresponsive while bzr contemplated
busily. I gave up at that point.

bzr may be fast enough in a POSIX world, but on Windows the native
package is slow. Mercurial is also written mostly in Python and is
blindingly fast by comparison, so it isn't Windows or python that's
causing the lack of performance.

    AndyM