Re: RFS: jabbin
> Hello,
> On Tue, 29 Aug 2006, Andrew Donnellan wrote:
>>> On 8/25/06, Hubert Chan <hubert@uhoreg.ca> wrote:
>> So what would be the alternative? Would we have to use a makefile hack
>> (e.g. test for the existence of /dev/random and set a compile flag) or
>> something to figure out what random seed we should use?
Usually, you would do a test in ./configure.
> Unless you need cryptographically secure random-ness (which is
> unlikely since this is a test suite) any pseudo-random generator
> should be adequate as a source of randomness. So you could use (for
> example) some post-header portion of a gzipped file. Of course, this
> gives you "deterministic random-ness" which sounds like an
> oxymoron---but isn't according to recent results in Complexity!
I haven't actually looked at the source. If this is just a test suite,
and it isn't actually used for anything important, then it may be
sufficient to just use the output of time(2). I really don't know why
they're using rdtsc as a source of randomness...
It's probably not a good idea to use a portion of a known file, since
then your seed is completely deterministic, and you lose the advantage
of setting a seed.
>> Also would it be OK to include this as a Debian patch, esp. because at
>> the moment Debian doesn't have any non-glibc ports?
IMHO, if you include this as a Debian-only patch, then it would be best
to get rid of the whole #if section, and put in the "this should work
for all Debian ports" code, instead of having an #if test that tests the
wrong thing. But that's just IMHO.
But I think that upstream probably would want to have a more general
fix. And IMHO, we should always try to come up with general fixes,
instead of just Debian-specific fixes.
--
Hubert Chan - email & Jabber: hubert@uhoreg.ca - http://www.uhoreg.ca/
PGP/GnuPG key: 1024D/124B61FA (Key available at wwwkeys.pgp.net)
Fingerprint: 96C5 012F 5F74 A5F7 1FF7 5291 AF29 C719 124B 61FA
Reply to: