Re: [gnugo-devel] persistent reading cache inconsistency patch

[Gunnar Farneback]

Evan wrote:
> I renamed attack() and find_defense(), and added a parameter to control
> whether they check the persistent cache before doing reading.  attack()
> and find_defense() as before now just call the new functions.  When
> attack_and_defend finds an inconsistency, it tries calling attack() and
> find_defense() without caching to fix it.  This of course slows down
> attack() and find_defense by a small amount, but since those aren't the
> recursive functions, I don't think this is a big deal.

I don't like this patch. A nicer way to solve the problem would be to
add a method to zap a particular entry in the persistent cache, call
this when an inconsistency has been detected, and then redo the
reading.

/Gunnar