[Octave-bug-tracker] [bug #58800] BIST for rng sometimes fails

[Markus Mützel]

Update of bug #58800 (project octave):

                  Status:               Confirmed => Patch Submitted        

    _______________________________________________________

Follow-up Comment #17:

Thank you for the review.
I agree that it is not necessary to pass the lower bound to the
uniform_int_distribution constructor.

I also agree that it is enough to add only one element from the
non-deterministic source to the initialization vector to ensure that the seed
changes.
While Octave's random number generator is probably not meant for encryption
purposes, imho we should nevertheless attempt to make it more "secure" if it
is easily possible.
The result of random_devices ()-operator is an unsigned int. That might be
just 16 bit worth of "entropy" depending on the implementation.
I'm not an expert. But that feels like it might make it easier to "guess" the
state of the rng.

What about the attached change that - if possible - adds at least 32bit worth
of "entropy" to the initialization vector (which isn't excessive imho)? I
don't know if we should try to add a bit more.

(file #49539)
    _______________________________________________________

Additional Item Attachment:

File name: bug58800_random_device_v3.patch Size:2 KB
   
<https://file.savannah.gnu.org/file/bug58800_random_device_v3.patch?file_id=49539>



    _______________________________________________________

Reply to this item at:

  <https://savannah.gnu.org/bugs/?58800>

_______________________________________________
  Message sent via Savannah
  https://savannah.gnu.org/