| |
| |
| |
|
Page: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32
Comments:
<0> well, since I don't care which one gets decreased, I dopn't care how random it is
<1> It says you should be able to do it either way I believe.
<2> it says: Later in your app: $self->cfg('HTML_ROOT_DIR')
<3> centosian: your point about hash functions and randomness is very important if a little misplaced in this crowd.
<0> PerlJam: I know exactly what I need
<0> aand that is to decrease the value of one of my values... any one. I really don't care which
<4> perljam: well thanks I guess :)
<1> but under the new() method docs, it says you can use PARAMS => {hashref}
<2> tomkp: right, so you need proper randomness.
<0> integral: explain
<3> tomkp: that's real randomness that you want.
<1> Look at the box right above that one
<2> godzirra: yes, but you access them with $self->cfg
<1> Ah.
<1> Duh.
<2> godzirra: it does not mention using ->params()
<1> Got it, thank you.
<3> well, not *real* randomness, but as close to real as we can get on a computer anyway
<0> I think I understand why you say that
<0> but I don't think I really do
<2> tomkp: can you ever do this twice in a row without modifying the hash in between?
<2> (if yes, this is the trivial case :-)
<0> no
<0> it will only ever be done once
<5> There Can Be Only One.
<2> PerlJam: you can get these wee PCI boards that generate proper random numbers
<0> if it needs to be done again, the haseh will be regenerated
<5> that's a weird, untestable algorithm
<6> P::C::I boards?
<5> I hate weird things like that
<2> ah! well if you generate the exact same hash you'll get the same answer
<5> integral - nope
<5> thanks to changes in 5.8
<2> merlyn: no?
<0> it really doesn't matter
<7> PerlJam: We can get TRULY random numbers from a computer...
<2> merlyn: random hash function for *each* hash? to provide security for long running apps?
<5> it was a denial-of-service attack vector
<3> centosian: A friend of mine was an expert witness in a court case involving the non-randomness of a the jury-selection process in the county I live in. It turns out that the jury selection software was making a hash based on zip code, person's name, and maybe a couple of other things. So much for "random" jury selection. :-)
<5> integral - yes
<0> because if the valu has gone to 0, that key will not be included next time the hash is created
<8> Hi PerlJam
<2> cool!
<5> you can turn it off with some pragma
<7> PerlJam: http://www.idquantique.com/products/quantis.htm
<2> merlyn: env var I thought
<4> perljam: like random airport screening
<5> but by default, my %x = (a => 1, b => 2) will be roughly 50/50 ordered
<2> merlyn: PERL_HASH_SEED specifically
<0> and if that value is still more than 0, then it doesn't matter if the same key is selected
<5> there ya are
<0> honsetly... it don't matter!!
<7> I always wondered how successfully IRC could be used for random number generation.
<2> tomkp: the other proiblem is that you'll get different stuff happening depending on perl version.
<5> there's a lot of entropy in newbie questions
<2> tomkp: right, so just put the shuffle in :)
<4> yaakov: poorly. I prefer using lavalamps
<5> actually, just watching pastebin might do it. :)
<9> what's a good hourly rate for a systems engineer with 8 years good exp ?
<2> merlyn: they end up like buses though
<5> how many years of bad experience?
<3> merlyn: no way, they're always asking the same stuff over and over again.
<9> on a 12mo contract to hire
<4> leku: more
<9> no bad exp
<9> been working for 10
<5> oh - not as much
<5> you have to get scars to be worth more
<10> leku: 1000$/h
<9> i'm thinking 70/hr
<10> that's "good" ;)
<0> if I don't care which key is selected, why would I care which one might be selected?!
<4> leku: WHERE?
<5> yeah, $1000/hr that'd be a great rate
<11> Yaakov: I did it. I managed to get a stddev about 20% higher than /dev/urandom on Linux with no PRNG hardware by running 5 warez channels on efnet through a fairly decent algorithm (imho).
<3> leku: reasonable (depending on where you're at)
<9> i would have to cover my own insurance, me + child
<5> Oh - you mean a rate you could GET?
<9> yeah
<0> as long as one gets chosen, thats fine
<6> It really depends on the market you are in, I'd say.
<4> and the cost of living in that region
<7> infi: I suspected that interval timing might be a way to do it.
<2> infi: umm, /dev/urandom isn't as good as /dev/random...
<11> Yaakov: Still needs a bit of work. I tend to get higher-than-average percentages of "random" characters at the characters that coincide with powers of 2.
<3> tomkp: do you happen to know one of the keys? If so, just choose that one all the time.
<9> 70 is about 134k a year
<6> Manhattan, $70/hour seems reasonable.
<10> I agree. There are markets that consider such wages an insult, either for being too high or too low
<0> PerlJam: no, I don't
<2> /dev/urandom will generate data using pseudo-random numbers when the entropy pool is empty
<11> integral: I know, but generating a full megabyte of data from /dev/random for comparison is time-consuming :)
<5> 70/hr as a contractor is not 134k a year
<2> entropy is slow ;-)
<11> yes, it is. :|
<6> standard = 2000 hours = 140K
<6> 50 weeks by 40 hours
<7> infi: Can you dispose of them?
<5> the rule of thumb for contract vs salary is that $N/hr = $Nk/yr
<4> yeah but contractors specd a lot of time *looking for work* typically
<5> because as a contractor, you have a lot of overhead and downtime.
<4> spend
<9> so
<2> and less paid holiday presumably
<7> infi: If you just discard that less random data...
<9> asking for 70$/hr is not bad?
<5> if you were paid $50K/yr as a salaried guy, $50/hr is about right
<3> leku: yes, $70/hour is fine.
<9> cool
<5> so look at the salaries in the industry
<0> anyway: (keys %$subscriber_tickets)[0] does not seem to work!
<9> i'm making about 75 now
<7> infi: No, nevermind, that would leave a non-random hole.
<9> so should I ask for more ?
<5> tomkp - it gives you a random key
<4> leku: ask for exactly what you want
<5> what part of that "does not work"
<11> Yaakov: I already do, I run the generated hashes through Compress::Zlib, then run the Bell Test (2 consecutive bits: 0+0 = discard, 0+1 = 0, 1+0 = 1, 1+1 = discard) to discard long runs of bits
<0> Compilation failed: syntax error at (eval 886) line 201, near ")["
<2> eval: [map { my %h = (1 => 1, 2 => 2, 3 => 3); (keys %h)[0] }, 1..5]
<12> integral: Error: syntax error at (eval 142) line 1, near "},"
<2> eval: [map { my %h = (1 => 1, 2 => 2, 3 => 3); (keys %h)[0] } 1..5]
<5> tomkp - you got something else going on then.
<12> integral: Return: ['1','1','1','1','1']
<11> Yaakov: I am going to try a different compression algorithm, haven't gotten around to it yet. I was pretty impressed with the results so far, though. let me see if I still have some of the data lying around.
<7> infi: But that just makes it LOOK random to you, it doesn't mean it IS random!
<5> yeah, that's always going to return the same thing
<3> tomkp: show more of the surrounding code.
<2> long runs are not random?
<13> infi: er, shouldn't you be compressing first and hashing second?
<11> Yaakov: I'll show you the output from the statistical generation
<4> I'm writing a program to return a list of 1's in random order
<4> it's hard to test
<0> my $ticket;
<0> die (keys %$subscriber_tickets)[0];
<0> if (sum values %$subscriber_tickets > 0) {
<5> eval:[map { my %h = (1..10); (keys %h)[0] } 1..10 ]
<12> merlyn: Return: ['1','1','1','1','1','1','1','1','1','1']
<5> eww.
<7> infi: What I mean is, there are long runs in random data, if they happen.
<10> eval: (split/,/,"foo,bar")[0]
<12> waiting: Return: foo
<2> tomkp: put a + between the die and the +
<5> not very random
<5> oh - maybe buubot isn't running 5.8.3
<5> eval $]
<5> eval: $]
<12> merlyn: Return: 5.008007
<5> odd
<0> integral: a +??
<5> I wonder if that's turned off
Return to
#perl
or
Go to some related
logs:
#ldap
python dump listctrl
#web
#lisp
+Access denied +debian-sys-maint +Ubuntu
parent.framename firefox
php 404 error page gentoo
xmms goodnight halt
ubuntu quit perl
xen-sources overlay
|
|