Re: [gnugo-devel] Owl persistent cache

[Daniel Bump]

> This is fixed by the appended patch, after which we instead get the
> much more reasonable (but possibly too small) active area below.

This patch is a HUGE speedup. I played two games versus 3.0
before and after the patch. The first game had 478 persistent
owl cache hits. The second had 1592 hits, and (surprise!) played
with acceptable speed on this fairly slow old Pentium 2.

This means that we can contemplate releasing 3.2 soon, say
around the end of March.

I hope to release 3.1.27 tomorrow morning and will send a further
message about 3.2.

Gunnar is planning one slightly radical patch implementing
undo and cleaning up the interface/ directory but apart from
this we hope that as we enter the late phase of the release
cycle we can concentrate more on stabilizing patches, cleanup
and other things. During the late phase of the release cycle
owl tuning and radical patches which risk breaking things or
making the code slower will be frowned on.

Dan