krita random_generator failure

[Matthew Woehlke]

(Apologies to Boudewijn for not replying to his mail, but I seem to have 
misplaced it...)

With the latest change, KisRandomGeneratorTest::testEvolution fails 
(i.e. what is currently in the sources), because the PRNG happens to 
spit out the same value for two "nearby" pixels.

Since producing the same number twice in a row is no less random than 
any other number, I'm not too sure how concerned I want to be about 
this, but it made me check the same seed against the previous 
algorithm... and, well, the results of the old algorithm are clearly 
unacceptable from a visual standpoint (obvious artifacts, obvious pattern).

So... we either need a better tweak, a more permissive test, or... 
something I haven't thought of :-).

Cyrille suggested a spot where I could check in my test program; 
unfortunately I am somewhat swamped with $DAYJOB at the moment, but 
under the circumstances I will try to get to it tomorrow.

-- 
Matthew
Please do not quote my e-mail address unobfuscated in message bodies.
-- 
With sufficient thrust, pigs fly just fine. -- RFC 1925