You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Original bug ID: 7073 Reporter:@gasche Status: acknowledged (set by @damiendoligez on 2016-02-10T16:53:25Z) Resolution: open Priority: low Severity: feature Target version: later Category: standard library Tags: testsuite
Bug description
In the previous weeks many of the Pseudo-Random Number Generators (PRNG) in Javascript implementation were found to be much weaker than they should -- even at non-security, non-crypto applications, I am not talking about crypto-randomness here. See the following article on LWN.net for a summary of the issues found in Chrome, Firefox and Safari:
(This link is made available from my LWN subscription. Consider subscribing to support this excellent source of technical journalism!)
The PRNG of web browsers were all found insufficient by Pierre L'Ecuyer's TestU01 set of statistical tests, which seems to be the modern version of the venerable Diehard PRNG testsuite.
It could be interesting to test OCaml's random generator against this new testsuite. Our implementation is non-standard, partly because of OCaml's integger tagging which makes the output domain of our PRNGs non-standard.
Note: the web browsers decided to move to a PRNG named xorshift128+, found at
If you're interested in a better/faster PRNG, and you use a different number of tagging bits, I can put up together a similar generator with a different bit size.
This issue has been open one year with no activity. Consequently, it is being marked with the "stale" label. What this means is that the issue will be automatically closed in 30 days unless more comments are added or the "stale" label is removed. Comments that provide new information on the issue are especially welcome: is it still reproducible? did it appear in other contexts? how critical is it? etc.
Anyone is welcome to test with their statistical test of choice and report the results. I'll stick with dieharder, because I find it to be quite complete.
The conclusion is that the Random PRNG is solid, statistically speaking. Its only issue is that the API is missing a few operations, first and foremost splitting.
Original bug ID: 7073
Reporter: @gasche
Status: acknowledged (set by @damiendoligez on 2016-02-10T16:53:25Z)
Resolution: open
Priority: low
Severity: feature
Target version: later
Category: standard library
Tags: testsuite
Bug description
In the previous weeks many of the Pseudo-Random Number Generators (PRNG) in Javascript implementation were found to be much weaker than they should -- even at non-security, non-crypto applications, I am not talking about crypto-randomness here. See the following article on LWN.net for a summary of the issues found in Chrome, Firefox and Safari:
http://lwn.net/SubscriberLink/666407/44be03d386807580/
(This link is made available from my LWN subscription. Consider subscribing to support this excellent source of technical journalism!)
The PRNG of web browsers were all found insufficient by Pierre L'Ecuyer's TestU01 set of statistical tests, which seems to be the modern version of the venerable Diehard PRNG testsuite.
http://simul.iro.umontreal.ca/testu01/tu01.html
It could be interesting to test OCaml's random generator against this new testsuite. Our implementation is non-standard, partly because of OCaml's integger tagging which makes the output domain of our PRNGs non-standard.
Note: the web browsers decided to move to a PRNG named xorshift128+, found at
http://xorshift.di.unimi.it/
The text was updated successfully, but these errors were encountered: