> -----Original Message-----
> From: Sergei Golubchik [mailto:serg@stripped]
> Sent: Friday, October 17, 2008 10:16 AM
> To: Vladislav Vaintroub
> Cc: Kevin Lewis; commits@stripped
> Subject: Re: bzr commit into mysql-6.0-falcon-team branch (klewis:2865)
> Bug#39672
>
>
> Just a comment.
>
> I use rand quite a lot in the unit tests. It doesn't make them
> non-predictable, when few hundred threads repeat some operation
> concurrently few thousand times, statistically result will be the same,
> rand or not. I use my_rnd(), though, not system rand().
I see you point. Right, I should not have generalized. In high-concurrency
tests, with many threads , where scheduling is unpredictable anyway,
randomness does not matter and can be beaten statistically. This particular
case however is different. The thing is a bitmap, for which bit set and bit
clear or nextset do not work for some values of bit. Randomness can easily
end up in "can't repro" situations. I think more efficient method is that -
since the author knows the implementation, he can analyze which corner
cases it has and test for each of them specifically.
> Regards / Mit vielen Grüßen,
> Sergei
>
> --
> __ ___ ___ ____ __
> / |/ /_ __/ __/ __ \/ / Sergei Golubchik <serg@stripped>
> / /|_/ / // /\ \/ /_/ / /__ Principal Software Engineer/Server
> Architect
> /_/ /_/\_, /___/\___\_\___/ Sun Microsystems GmbH, HRB München 161028
> <___/ Sonnenallee 1, 85551 Kirchheim-
> Heimstetten
> Geschäftsführer: Thomas Schroeder, Wolfgang Engels, Dr. Roland Boemer
> Vorsitzender des Aufsichtsrates: Martin Häring
