On 09/14/2015 02:23 PM, Thierry wrote: > Hi, > > i am pretty remote right now (for a Sage tutorial actually;), but you can > count me in, starting from the next week. > > Some few ideas i planned to work on for a while: > > - random testing : the current random doctests uses a single seed, so we do > not get any more information when the doctests are run. We could > also have some truly random tests that, for example, test the > associativity in groups from random_elements. The more they are run, the > stronger Sage is. Then we could have som VM that only runs those random > tests while changing the seed each time. >
I've been meaning to bring this up. We need to get rid of the fixed seed. I know for sure I've written random tests in the past, not knowing that sage silently replaces its random number generator with https://xkcd.com/221/. Unless you happen to stumble upon the "Library API" section of http://doc.sagemath.org/html/en/reference/misc/sage/misc/randstate.html the fact that your random numbers have been subverted will not be apparent unless you find some bug in your code that should have been doctested. The rationale given is that we need to be able to reproduce failing doctests. Well, why not just spit out the random seed at the beginning of `sage -t` then? See `random_cone()` for an example where I call set_random_seed() a million times. Had I not found a few bugs in the code that should have been tested, I probably would not have noticed that randomness was broken, and the function might have been merged with bugs. -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To post to this group, send email to sage-devel@googlegroups.com. Visit this group at http://groups.google.com/group/sage-devel. For more options, visit https://groups.google.com/d/optout.
