Mirror of forum.nim-lang.org

1110 :: Is epochtime() the Nim equivalent of time(NULL) in C?

• jarod (orginal) [2015-04-08T09:54:50+02:00] view original

  time(NULL) is frequently used to seed C's RNG, I'm doing my own experiment with seeding a personal RNG. Is epochtime() the Nim equivalent? Checking the documentation and some stack overflow questions seems to suggest they are. However I'd thought I'd seek clarification.

---

• Araq (orginal) [2015-04-08T13:58:30+02:00] view original

  Nim's equivalent is math.randomize but don't use that because it doesn't produce random numbers that are random enough to be useful enough to fly to the moon with them which is obviously what every programmer wants to do.

---

• BlaXpirit (orginal) [2015-04-08T23:37:58+02:00] view original

  Yes, epochTime is time(NULL). Yes, some use it to seed RNGs. But it is a horrible idea.

  Please use urandom (external library) to obtain bytes for seeding. You can also get a random integer:

  import random.urandom
  
  var sysrand = initSystemRandom()
  seed(sysrand.randomInt(uint32))
  Run

  You could also build your RNG into the framework of that library to not reinvent utility functions on top of it.

---

• Jehan (orginal) [2015-04-09T01:13:27+02:00] view original

  BlaXpirit: But it is a horrible idea

  It is not a horrible idea if you don't need it for cryptographic purposes (for non-cryptographic purposes, urandom is overkill) and it avoids an external dependency (not to mention a dependency on /dev/urandom, which may not always be available, e.g. inside a chroot jail or on some HPC architectures). Especially if, as the OP seems to indicate, he is already writing an RNG of his own (for whatever purposes, e.g. because he needs it to have certain mathematical properties).

---