| |
| |
| |
|
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> infi: oh, add "given N different samples of your random data" to that :)
<1> oh, heh
<0> Stormbreaker: why wouldn't it?
<2> Stormbreaker: Because it does the same thing only one whitespace character at a time. Mine removes whole runs of whitespace all at once. It's a minor thing.
<3> it's a hypothesis test, you cannot eliminate false positives or negatives, just minimise them.
<1> hobbs: care to see the hashdata sub I'm using? I misremembered, I do actually do the compression first.
<2> Am I the only one who finds it odd that people are discussing mathematics of knuthian proportions on #perl? ;-)
<0> infi: sure
<0> PerlJam: only very minor knuthian proportions ;)
<4> "knuth jr", as it were
<2> heh
<0> besides, Perl _puts_ the Art in Computer Science!
<5> "The very fact that it's possible to write messy programs in Perl is also what makes it possible to write programs that are cleaner in Perl than they could ever be in a language that attempts to enforce cleanliness. The potential for greater good goes right along with the potential for greater evil." -- Larry Wall
<6> perl is like the Force[tm] :)
<7> perl IS the force
<6> heh
<0> BinGOs: that already followed by the transitive property
<8> the force is strong with perl yesss.
<9> half the world is run by perl
<0> everyone knows Perl is like duct tape
<9> the other half ****s
<10> "infi" at 219.120.193.247 pasted "hashdata() and belltest() for hobbs" (60 lines, 1.7K) at http://sial.org/pbot/17442
<0> and everyone knows the force is like duct tape
<8> it has a light and dark side and holds the univers together
<4> dark side of perl = golf + japhs
<6> with a chewy mint centre.
<4> I wish golf and japhs and obfuperl didn't exist
<8> perlbot, force
<1> hobbs: oops, the input data is... just the message from the user.
<6> speaking of the dark side, here's merlyn :)
<0> infi: hmm.... improvement comes to mind ;)
<8> can i add words to perlbot?
<6> can you ?
<4> any particular words?
<4> word of the day, perhaps?
<1> hobbs: indeed. I know it is vulnerable to a replay attack offhand.
<6> GumbyBRAIN: can RiXtEr add words to perlbot ?
<11> Yeah - i know, i talk to perlbot in the words?
<9> merlyn: are those perl sourcecode obfuscators?
<0> infi: maintain your zlib object between calls to hashdata, and if it doesn't produce any data on a given run you don't worry about it -- you'll get it next time around
<6> Guess so.
<8> GumbyBRAIN, force
<11> It might force a scalar ref of an irregular wave function, integral?
<6> perlbot: help > RiXtEr
<4> perlbot, equable is "Equal and uniform; not varying." (from http://dictionary.reference.com/wordoftheday/)
<12> added equable to the database
<4> heh
<13> waiting: no a JAPH is a Just Another Perl Hacker.
<0> infi: only ever taking the very early bits of the gzip stream is hurting you pretty badly, I expect.
<4> I need a bot to feed perlbot now
<14> I want to print _$ + \n, but how?
<14> print($_);
<4> print "$_\n";
<6> print $_, "\n";
<1> hobbs: oh, the gzip header... right.
<13> waiting: it's a small Perl program that people put in signatures that says something smart, while its code being unconventional.
<2> print $_ . "\n"; # for completeness ;-)
<0> infi: well, you throw that out, but still. It takes a little bit for the algorithm to hit its stride
<2> print $_; print "\n";
<4> perlbot, japh is http://en.wikipedia.org/wiki/Just_another_Perl_hacker
<12> I already have an entry for japh
<4> no, perlbot, japh is http://en.wikipedia.org/wiki/Just_another_Perl_hacker
<9> rindolf: ah like perl -e "super;duper;execute(me(please)%$!"
<8> what do i need to add at the end of a line to do a cr?
<4> hmm. did that work?
<15> printf "%s\n", $_; print sprintf("%s\n", $); ...
<4> how do you update perlbot?
<1> hobbs: I haven't worked on this in a few weeks, you and someone else recommended another algorithm a couple of weeks ago. Shall I give that a try instead?
<0> infi: it's a well-known fact that practical compression algorithms work better on bigger blocks of input data, and for your purposes more compression == less overestimation of entropy
<6> perlbot: help > merlyn
<9> I never do that :S
<4> heh. "oh, turtleneck Phrase Jar!"
<3> rzip's meant to be very good, but very slow
<0> infi: depends what the other algorithm is. Yours isn't bad -- just very slightly broken ;)
<4> I remember the glee I had when I found that anagram.
<1> hobbs: ok, then, should I maintain the object, keep a large stream, not generate any data until the stream is up over some arbitrary size, and then start pulling out bytes from the head end at a roughly equal speed to how quickly they are arriving?
<4> perlbot, relearn japh is http://en.wikipedia.org/wiki/Just_another_Perl_hacker
<12> relearned entry for japh
<16> infi: Isn't bzip2 really good at compression?
<0> infi: you don't need to worry about that. Just feed the data to zlib and don't ever force the stream to flush -- take data whenever it feels like giving it to you
<1> centosian: yes, but slow.
<16> infi: So you want something almost as good but faster?
<17> What's the P::C::I command to get the list of the room people? Something like a list of hashrefs like { jsn => 'u=jsn!$@me.com' } returned would be nice
<16> for some definition of 'almost'
<6> jsn: you using vanilla POE::Component::IRC ?
<0> infi: so it's asynchronous, right? You p*** data in as you get it, and you get data out of zlib when _it_ has some for you
<13> The cvs client to my workplace's CVS server hangs.
<17> BinGOs, pretty much, yeah.
<1> hobbs: yes, I see what you mean.
<13> I hate, detest, loath, abhor and despise CVS.
<0> infi: and that will ensure you get good compressed data
<2> rindolf: install subversion!
<8> how can i intigrate a nice value to a program?
<13> PerlJam: I have Subversion installed.
<2> rindolf: sense making none you are.
<6> jsn: the irc command to get the people in a 'room' ( we call it a 'channel' ) is 'who'
<16> rindolf: but is it better than no versioning? :)
<1> hobbs: this may substantially reduce my hits on the 2^x byte values, too. good idea
<2> er, RiXtEr even
<3> RiXtEr: look in POSIX
<13> PerlJam: but everybody else are using CVS.
<2> rindolf: so convert them! :)
<13> PerlJam: I'd better write a let's switch to svn memo.
<3> POSXI::nice specifically
<8> PerlJam, why didn't that make sense?
<6> jsn: but if you want the component to deal with all that channel tracking **** for you, replace POE::Component::IRC in your code with POE::Component::IRC::State
<2> RiXtEr: perhaps because I wasn't paying enough attention.
<1> hobbs: aside from that, do you see any major flaws (aside from the obvious aforementioned replay attack vulnerability)?
<8> PerlJam, :)
<17> hrm, i'll look into it
<2> rindolf: Or you could start building work processes that rely on on subversion and present it as a fait accompli
<2> :-)
<6> jsn: poco-irc-state comes with a lot of shiny methods for querying who is on channels the bot is on, etc.
<13> centosian: CVS > patches and tar files >> no versioning. But Subversion >>> CVS.
<0> infi: no, seems alright. And if you use the compression properly, and with the bell test, your entropy estimate should be _about_ on the mark. It still probably needs a bit of revision downwards, but I don't have any thoughts at the moment on how to do it
<0> infi: I'd suggest trying the bit with the compression and see how it affects your results
<1> hobbs: nice, thanks for the input. I will try tweaking with the compression, and perhaps a couple of other compression algorithms entirely.
<8> integral, man perldoc POSIX and perldoc -m POSIX **** for telling about nice :) would i do... use POSIX::nice; nice(+10);
<18> POSIX. To access this perldoc please type, at a command line, 'perldoc POSIX'. You may also find it at http://perldoc.perl.org/POSIX.html
<0> infi: if bzip2 isn't too slow for your machine to keep up, I'd give it a spin. The perl module has an interface that makes it a drop-in replacement for zlib after changing a few function names.
<16> rindolf: someday I'll have to migrate my cvs repos to svn
<3> RiXtEr: pardon? perldoc POSIX and search for nice. it is the first hit.
<0> centosian: svn is pretty enjoyable to work with
<3> and try thinking. use POSIX::nice; is wrong. I don['t know where you learnt it but go and unlearn it.
<1> hobbs: bzip2 won't really be too slow. the warez channels do get a bit rollicking during evening hours US time, though ;)
<1> it really is a fairly low amount of input data, though, so I imagine it'll be fine.
<13> centosian: svn ownz!
<16> since I don't typically branch I think I avoid some of the CVS probs. Is it painful to migrate from cvs?
<2> centosian: depends
<3> centosian: with subversion one can *rename* files!
<16> WHOOT!!++
<5> LIES
<2> centosian: how much of your work flow depends on certain features of CVS?
<13> centosian: use cvs2svn
<16> PerlJam: um a decent amount. I do automate a lot of cvs opertation
<2> centosian: but, in general, it's *easy* (use the program that rindolf mentions)
<16> s/kasdjfkj/operations/
<0> centosian: not so much. svn explicitly tries to be "like cvs but better" instead of "the latest greatest paradigm in version control to reach this corner of the galaxy"
<2> centosian: for instance, CVS tracks *file* revisions, but subversion tracks *tree* revisions. So some things are "hard" in subversion.
<16> My one sticky point is sometimes I used to checkin binaries (gifs and such) without -bk, which I understand I need to resolve before moving to SVN or the files will breadk
<13> centosian: with Subversion branching is much more painless than with CVS.
<16> I'm gonnna look at cvs2svn
<2> centosian: you may not. svn is fairly smart about file types.
<16> somewhere I read that if you have binary files in CVS that weren't set to be treated as binary, that had to be fixed first (like I said above).. That was data I read like 18 months ago though if memory serves
<1> rolling back an update in the repository itself is a bit difficult. say someone commits like a gig of data, not considering the impact. you have to dump the repo, filter the revision out, and then restore :|
<0> centosian: I don't know about the details of the conversion process from CVS, but I expect any mistakes in that department could be fixed afterwards with just an "svn propset"
<1> that's been my only complaint about it so far.
<16> hobbs: At worst I'd just put the files into SVC anew
<1> though I don't know if CVS is similar in that respect, as I never had to do that.
<5> infi: you can feed urls to svn rm
Return to
#perl
or
Go to some related
logs:
tar relocation error debootstrap
netinstall fedora repomd.xml
#css
suse 10.1 divx codec rpm
gentoo HTTP request sent, awaiting response... 416 Requested Range Not Satisfiab
mondo-rescue gentoo unmask
lisp special forms defmacro if
#web
#perl
#perl
|
|