#perl
28 December 2007
Total 45 pages. You are browsing page 5/45.
First :: Prev :: [...] [3] [4] [5] [6] [7] [...] :: Next :: Last
02:28 <****> I only need the top x items (not many, about 5, but not the same each time)
02:28 <****> What will you do?
02:28 <****> Hello there
02:28 <****> TPC - what version of perl?
02:28 <****> is there a faster way to do it?
02:28 <****> TPC - use a heap
02:28 <****> Limbic_Region, 5.8.8
02:28 <****> Limbic_Region++
02:29 <****> right, that's a heap job
02:29 <****> ok, I'll google that, thanks
02:29 <****> TPC - I have a node on this at PerlMonks if you are interested
02:29 <****> search cpan
02:29 <****> Limbic_Region, sure
02:29 <****> fwiw, reverse sort is done in place in 5.10.0 (and so more memory efficient)
02:30 <****> http://perlmonks.org/?node_id=427029
02:30 <****> Heap::Simple is probably the way to go
02:30 <****> ok
02:31 <****> looks like 5.10 is in debian experimental, I could just install it
02:32 <****> or maybe not
02:32 <****> I have a question about http and browser timeout. It appears that if a browser successfully sends it's HTTP request to the server it will wait what seems like *forever* for the server to either close the connection or send something back.
02:32 <****> TPC - you missed an important point
02:32 <****> memory efficient
02:32 <****> it isn't going to do a lot in the way of speed
02:33 <****> if you are only interested in the top N items, sorting the entire list is probably a waste
02:33 <****> it all depends on your specific situation of course
02:33 <****> Read the thread I posted, quite enlightening
02:33 <****> I learned a lot from it anyway
02:33 <****> I was just watching a open HTTP connection to a server and the browser stayed around for many minutes waiting for a response.
02:33 <****> Given that, why do so many people seem to need to do keep-alive things?
02:34 <****> Is that just to pacify Apache?
02:34 <****> are you sure your browser isn't using that?
02:35 * Limbic_Region is confused by jjore-m's question
02:35 <****> http is a stateless protocol
02:35 <****> so unless we are talking about stuff at the transport layer
02:35 <****> Limbic_Region, yah, I will read it
02:35 <****> I'm sure my browser successfully connected, that it sent a http request, that then *nothing* happened afterward. The server didn't have to send anything back to the client to indicate that it was still thinking.
02:35 <****> I am not sure
02:35 <****> Limbic_Region, however memory usage is not really an issue here
02:35 <****> Just not sending anything seemed to be sufficient.
02:36 <****> TPC - a heap will be fast, sorting the entire array may not be
02:36 <****> TPC - I was just commenting that there was an optimization in perl 5.10.0 to do reverse sort array in place
02:37 <****> ok
02:37 <****> does the whole :utf8/:encoding(utf8) thing work when sysread is being used? :|
02:37 <****> So I'm just confused about what problem people are solving in the "heartbeat" issue on http://fit.c2.com/wiki.cgi?PreventingBrowserTimeout.
02:37 <****> jjore-m: but there are many different browsers and servers, you don't know if that will be the case with all of them
02:38 <****> jjore-m: oh and it might be long but it DOES eventually timeout :)
02:39 <****> someone in #mysql ran into that problem the other day when they were trying to load a .sql file
02:39 <****> TPC - see Heap::Simple::XS if you can compile
02:39 <****> How long is long, Khisanth?
02:40 <****> I have also ran into that on our own server, apache was perfectly fine with it and there was never a timeout but on the live site the connection goes through a load balancer and that will kill the connection if nothing is sent after a while
02:41 <****> Ah, hmm.
02:41 <****> jjore-m - I am sorry for being so dense, but are you talking about a TCP timeout?
02:41 <****> oh wait
02:41 <****> I get it
02:41 <****> No, http response timeout.
02:43 <****> along the lines of load balancers - it may also need to be used for proxy affinity
02:43 <****> IE - let's say your corporate network has you go out through a squid farm of a dozen proxies
02:44 <****> despite http being a stateless protocol, you will stay attached to the same proxy to maintain the illusion of state
02:44 <****> if the proxy thinks the session has ended, you may end up on another proxy
02:44 <****> the keep alive can be used to make sure that doesn't happen
02:45 <****> those both make sense, sure. ok.
02:46 * CPAN upload: selfvars-0.04 by AUDREYT
02:46 * CPAN upload: Perl6-Doc-0.32_09 by LICHTKIND
02:47 * luke-jr ($*)@$
02:47 * jjore-m wonders if audrey figured out her reifying other people's @_ problem for that selfvars.
02:48 <****> no one reifies my @_ but me
02:49 <****> jjore-m - which reify problem is that?
02:49 <****> the source is quite short
02:49 <****> I was just checking it out
02:50 * f00li5h meows
02:50 <****> What does it do?
02:50 <****> buu: meow'ning
02:50 <****> how's things?
02:50 <****> there was something like "shift off your caller's @_" which requires that your caller's @_ get reified first.
02:51 <****> buu - gives you shortcuts to $self and @args in OO modules without you having to do my $self = shift; type stuff
02:51 <****> Oh.
02:51 <****> Does it use the obvious way or can it create lexicals somehow?
02:52 <****> they are tied
02:52 <****> Hrm.
02:52 <****> oh no.
02:53 <****> The CPAN near me doesn't show version 0.04 yet
02:53 <****> can't check the change log
02:53 * luke-jr <-- pissed
02:54 <****> apparently, you were swearing just a minute ago
02:54 <****> [20:50] * luke-jr ($*)@$
02:54 <****> I thought you had mistyped your prototype at first
02:54 <****> ;-)
02:54 <****> meh
02:54 * luke-jr stabs Perl ?
02:55 * luke-jr begins to ponder porting to C
02:55 <****> gosh.
02:55 <****> Limbic_Region: i guessed it wsa a regex...
02:56 <****> luke-jr - is your issue the encoding or speed?
02:56 <****> encoding
02:56 <****> or rather, NOT encoding
02:57 <****> :raw ?
02:57 <****> everything I work with is already utf-8
02:57 <****> but Perl insists on treating it as Latin-1 somewhere
02:57 <****> where is Juerd when you need him
02:57 <****> the really annoying thing is that I had it working and it just stopped with no warning or changes to code ?
Total 45 pages. You are browsing page 5/45.
First :: Prev :: [...] [3] [4] [5] [6] [7] [...] :: Next :: Last