the bigg Posted April 16, 2009 Share Posted April 16, 2009 I use SCS(II) in my WeiDU testing suite, but need to have full reproducibility of installs to ensure meaningful results. Would changing RANDOM_SEED 3.14 to RANDOM_SEED 0 in the tp files be enough to remove the results of randomness, or would I have to alter the perl files as well? Link to comment
DavidW Posted April 17, 2009 Share Posted April 17, 2009 I use SCS(II) in my WeiDU testing suite, but need to have full reproducibility of installs to ensure meaningful results. Would changing RANDOM_SEED 3.14 to RANDOM_SEED 0 in the tp files be enough to remove the results of randomness, or would I have to alter the perl files as well? The Perl script doesn't do any randomising. I'm actually surprised RANDOM_SEED 3.14 doesn't ensure reproducibility too, though possibly I'm misunderstanding how WEIDU's randomiser works. (I have a feeling I put that in during my own testing to ensure reproducibility, and forgot to take it out.) Link to comment
the bigg Posted April 17, 2009 Author Share Posted April 17, 2009 Thanks for the answer. RANDOM_SEED will do a reproducible seed if you pass an integer (or an expression evaluating an integer), otherwise it'll be non-reproducible if you pass a string that doesn't evaluate to an integer (in particular, 3.14 falls in the latter cathegory). Link to comment
Recommended Posts
Archived
This topic is now archived and is closed to further replies.