Camelia, the Perl 6 bug

IRC log for #mojo, 2012-09-23

| Channels | #mojo index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:35 davido joined #mojo
00:42 doswaldcal joined #mojo
01:49 jwang joined #mojo
02:02 human39 joined #mojo
02:17 dross best caching strategies for perl, what are they?
02:17 dross Mojo::Plugin::Cache isn't sufficient, I need something else.
02:17 dross I was thinking about memcached, obviously it's widely used, but want to know if there are any others which are interesting
02:40 noganex_ joined #mojo
04:08 Adura memcached, tried and true, why not?
04:18 zpmorgan joined #mojo
04:58 dross Adura: I guess
04:59 dross Adura: I'm always doing things very specifically anyway
04:59 dross I know RoR and Django, while they make things easy, they also make complex systems difficult to accomplish
05:52 Foxcool joined #mojo
06:02 xaka joined #mojo
06:33 Vandal joined #mojo
06:57 dross joined #mojo
06:58 l3l1p joined #mojo
07:01 kvorg joined #mojo
07:07 rem_lex| joined #mojo
07:21 * dross yawns
07:23 l3l1p joined #mojo
07:32 d4rkie joined #mojo
07:43 Britzel joined #mojo
07:45 Htbaa joined #mojo
08:21 dross o.o
08:21 dross odd, I guess I've never ran in to the failraptor until today
08:21 dross twitter failwhale spoof I guess
08:38 Foxcool joined #mojo
09:38 dross :)
09:51 zpmorgan batman++, marcus++, Mojo::Redis seems to be working great!
09:56 zpmorgan The only thing I don't understand is exactly what makes subscriptions timeout. If unsubscribe works, maybe connections could stay alive by subscribing & unsubscribing to random channels with execute()
10:05 Foxcool joined #mojo
10:33 dbr joined #mojo
11:19 bpmedley left #mojo
11:27 stephan48 joined #mojo
11:39 batman joined #mojo
11:45 memowe joined #mojo
12:08 vishwanathms joined #mojo
12:08 vishwanathms_ joined #mojo
12:22 zpmorgan hi batman
12:29 sri hmm, intersting that socketpair() is so portable
12:30 kitt_vl joined #mojo
12:48 bpmedley joined #mojo
12:53 nic thanks sri & xaka for the notes about Child & socketpair
13:03 inokenty joined #mojo
13:21 sri btw. pipe() works on windows too
13:22 sri but you get fake handles there without file descriptors
13:22 sri so you can do perlio stuff, but no lower level syscalls
13:24 sri i've actually thought a bit about factoring out hypnotoad/morbo fork handling, but killing worker processes reliably is a bit of a pita
13:47 osprey_ ad
14:00 batman zpmorgan: hi. what up?
14:01 zpmorgan you mentioned timeouts & redis. was it mojo streams that were timing out?
14:01 zpmorgan batman++, btw. Mojo::Redis is nice.
14:02 batman yes. the stream has a default timeout on 10 second i thing
14:02 batman *think
14:03 zpmorgan I'm not too keen on the idea of having a subscription reconnect every 10 seconds
14:03 batman zpmorgan: neither am i. that's why i'm doing $redis->timeout(600);
14:03 zpmorgan maybe it could ping redis by subbing & unsubbing to dummy channels with execute.
14:03 batman we might want to discuss a higher default value for the timeout
14:04 asarch joined #mojo
14:04 zpmorgan I'd like for the connection to close whenever its websocket closes.
14:04 batman i don't think that works. the sub/pub channel is a different connection
14:04 zpmorgan http://i.imgur.com/lPggr.jpg :)
14:04 batman zpmorgan: that works. just let the redis object go out of scope and set a high default timeout when you start
14:06 zpmorgan okay, then I guess it could ping to its own channel..
14:06 batman yeah. but just set a high timeout and you should be fine about the reconnect
14:06 batman if nothing happens after 10 minutes, then you might want to close down the connection anyway...
14:07 zpmorgan unless it's 10 minutes of deep thinking about the next move :)
14:07 batman then set timeout(3600); :P
14:07 batman anyway - you need to handle reconnects anyway
14:08 batman disconnects i mean.
14:08 batman i might want to add that directly to Mojo::Redis
14:08 zpmorgan why? are connections unreliable?
14:08 zpmorgan I don't like the idea of missed messages
14:08 batman like $redis->auto_reconnect(1); or sometihng
14:08 batman yes. is that surprising to you?
14:08 batman networks suck.
14:09 batman that they work as good as it seems is an illusion :P
14:09 batman you won't get missed messages if you do it the right way.
14:09 zpmorgan it's not over a network in my case.
14:09 batman where is it? localhost?
14:09 zpmorgan yes
14:10 batman why are you using redis then? you should just write it in js :P
14:10 zpmorgan mainly for pub/sub between multiple workers
14:12 zpmorgan what's the right way? I a disconnection means a time when you can't receive messages.. How does that not mean missed messages?
14:13 batman maybe you should use a list instead...
14:16 batman i'm using pub/sub and lists. that way you can read from the list if there's something you've missed out on
14:16 batman but also get the data at once if you're "connected"
14:18 zpmorgan okay. It's just that planning on local redis connections disconnecting randomly seems a bit pathological
14:19 batman that doesn't (shouldn't) happen on localhost, but it may very well happen if you put your application elsewhere
14:20 batman weird stuff that may happen, always happen once you get enough traffic
14:21 sri counting on it not happening is just naive
14:21 batman exactly
14:22 batman been there. ate that hat :P
14:22 batman still there occasionally unfortunately...
14:23 sri better expect the worst case and run a chaosmonkey :)
14:23 batman but mostly because i haven't thought about things, not because i ignore them
14:23 batman :)
14:23 sri http://www.codinghorror.com/blog/2011/​04/working-with-the-chaos-monkey.html
14:25 batman sweet
14:27 Britzel_ joined #mojo
14:38 d4rkie joined #mojo
14:38 batman https://github.com/marcusramberg/m​ojo-redis/blob/master/t/unicode.t <-- is this a valid unicode test?
14:40 batman marcus: i was thinking about changing default timeout to 15 i Mojo::Redis to make it consistent with the default timeout in Mojolicious
14:44 sri finding the right timeouts is hard
14:45 batman yeah. i think i would set it to 300 or something, but it's better to have it consistent and low than high and random
14:46 biz joined #mojo
15:03 batman zpmorgan: would you find this useful: https://github.com/marcusramberg/mojo-redis/co​mmit/90846b0ad06e16c275c7aa4de16ad7a2045eb2ec ?
15:03 batman zpmorgan: btw: i've added close event to Mojo::Redis now (in master)
15:10 zpmorgan batman, maybe. I was only planning to use the most basic case.
15:10 batman "basic case" ?
15:11 zpmorgan er, I don't know.
15:11 batman ok :)
15:21 Foxcool joined #mojo
15:27 dross joined #mojo
15:54 abra joined #mojo
16:27 abra joined #mojo
16:28 jzawodn joined #mojo
17:41 xaka joined #mojo
19:44 mire joined #mojo
19:50 MidnighToker joined #mojo
20:14 mire joined #mojo
20:37 l3l1p joined #mojo
20:43 lukep joined #mojo
20:49 Adura Siri, find me an Auth plugin.
20:56 * dross yawns
20:57 dross Adura: google's voice recog just said mojolicious::plugin::authentication
21:46 Jason_ joined #mojo
22:03 Jason__ joined #mojo
22:12 mire joined #mojo
22:24 l3l1p joined #mojo
22:34 Britzel joined #mojo
22:47 l3l1p joined #mojo

| Channels | #mojo index | Today | | Search | Google Search | Plain-Text | summary