Trouble with 5150UUID-fix

goran.hultgren at bluefish.se goran.hultgren at bluefish.se
Fri Dec 20 09:01:57 UTC 2002


"Andreas Raab" <andreas.raab at gmx.de> wrote:
> Hi Guys,
> 
> Just by random chance I was running a VM that doesn't have the UUID
> primitives but which _was_ updated to 5156 and used UUIDs. I got a
> notifier the "#dev#urandom" does not exist and no way out. Looking at
> the code it seems strange to see how #makeUnixSeed works - first it does
> silently assume the presence of /dev/urandom and only resets the answer
> if it is curtailed (e.g., if the process is _terminated_ at which point
> resetting the answer is entirely pointless. So that makes me wonder if:
> a) the code really should check for running on a Unixy platform before
> trying to go to /dev/urandom
> b) the code really should use other means of testing for /dev/urandom
> than just opening the file
> c) if anyone on a *nix WITHOUT /dev/urandom has ever tried it...
> 
> Any ideas?!

Well,
	a) Yes - that might be a good idea. I assumed it behaved "correctly" if
the file wasn't there...
	b) Sure, perhaps that would be better than a)? I mean - if the file is
there it probably should work, right? Or we simply do both checks.
	c) I thought every *nix had urandom, but obviously that is not the
case.

Btw, Cees - why did you choose urandom instead of random? I read
somewhere that random ensures "good bits" and urandom doesn't but has
the advantage of not blocking. Not that it really matters but I am just
curious.

> Cheers,
>   - Andreas

regards, Göran




More information about the Squeak-dev mailing list