> > seq feature for other things as well, I think, provided it was well > > implemented - e.g. 'jot' seems to seed it's RNG from epoch seconds, > > which is no good, microseconds would be better. The disadvantage is > > Yes, but there's two points here: > 1) There's generally a way to the user to specify a seed. Tipically this > value is specified in the last argument (step value in nornal mode).
Would we have to keep BSD syntax? I think that's confusing - since specifying a seed would probably be rarely done, I think it should be a long option instead. (And I'm just saying that the default should be that if you run it several times in a row, each time it will give different results, which isn't the case if you use seconds. I guess perl seeds from /dev/urandom if available, and if not then some combination of time, $$, and something with memory allocation - but these are unimportant details) > 2) Today's implementations are using other randomizations than a simple > s?rand(3) -- take for example the OpenBSD's implementation (uses ARC4), > though this implemantion doesn't specify user-specified seed. I'm sure something better than the libc rand implementations wouldn't be objectionable, as long as it is actually better. > I'd sum the good decisions of these implementations up in seq this way > > 1) Generate 4 random values (ranging from 1 to 4): > $ seq -r 4 > > 2) Generate 4 random values (ranging from 4 to 7): > $ seq -r 4 7 > > 3) Generate 4 random values (ranging from 4 to 7, seed 3): > $ seq -r 4 3 7 > > 4) Generate an endless number of random values (useful with head): > $ seq -r > > Does anyone have suggestions on this? Hmm, I don't like the syntax. As I said, having the seed where 'increment' should be is confusing. And how do I generate 8 random values from 1 to 10000? How do I generate an endless number of random values from 4 to 7? Frederik -- ---------------------------------------------------------------- Frederik Eaton http://ofb.net/~frederik/ _______________________________________________ Bug-coreutils mailing list Bug-coreutils@gnu.org http://lists.gnu.org/mailman/listinfo/bug-coreutils