Re: [gnugo-devel] Crash again

[Gunnar Farneback]

Nando wrote:
> While doing some owl-tuning, I got this somewhere in nngs3.tst (around 840)
> :
> 
> ***assertion failure:
> reading.c:397 - !((acode != 0) && (dcode == 5 && dpos == 0)) near PASS***

Any chance to get a reproducible test case from this?

> The following patch solves the problem, but I haven't tried to
> figure out if it makes sense or not (to me, it seems strange that
> more than one attack on a string could be stored in the persistent
> cache). Possibly, the problem is somewhere else and/or "deeper" and
> this patch could be a bad solution.

As implemented it could happen that we get more than one hit in the
persistent cache. This has to do with the handling of the
remaining_depth field. The patch is probably fine.

/Gunnar