Camelia, the Perl 6 bug

IRC log for #mojo, 2011-07-23

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

All times shown according to UTC.

Time Nick Message
00:46 xaka left #mojo
00:47 perlrocks Twitter: "Mojolicious-Plugin-Bcrypt 0.03 by MINIMAL - http://metacpan.org/release/MINIMA​L/Mojolicious-Plugin-Bcrypt-0.03/" --cpan_new http://twitter.com/cpan_ne​w/status/94569157982564352
01:05 xaka joined #mojo
01:14 alnewkirk joined #mojo
01:20 alnewkirk is now known as anaio
02:04 tessio joined #mojo
02:47 tessio left #mojo
04:45 kaare joined #mojo
05:22 Kulag left #mojo
05:23 Kulag joined #mojo
05:54 robinsmidsrod left #mojo
05:54 robinsmidsrod joined #mojo
06:21 jwang left #mojo
06:28 xaka left #mojo
07:39 perlrocks Twitter: "Mojolicious:: HTML and Lite:: FillInForm:: Fill-in Form Lite - http://htn.to/8mcbws scrawl of Tatsumi" (ja) --yusukebe http://twitter.com/yusukeb​e/status/94672886241038336
07:40 perlrocks Twitter: "MVC development in Mojolicious + Xslate + Skinny - Dive into the Tech World! Http://htn.to/Duaya7" (ja) --yusukebe http://twitter.com/yusukeb​e/status/94673115581399040
07:40 anaio left #mojo
08:35 sri good morning fellow unicorns
08:36 * crab hastily hides his second horn
08:41 sri https://github.com/kraih/mojo/issues/184 # hmmmm, i'd like to hear more opinions about this
08:43 crab what's the problem? seems like an obvious improvement.
08:45 sri what woudl you use it for?
08:46 sugar joined #mojo
08:46 sri the question if it would be worth the maintenance cost
08:48 crab i have no use for it in my current work, but i know i have used multiple byteranges before. i can't remember why, though.
08:49 crab i wonder if it would be possible to write it as a plugin using an after_static_dispatch hook to transform the response into a multipart/byteranges if multiple byteranges are specified in the header.
08:50 crab but there are too many cases to handle comfortably.
08:52 crab it seems to me that either mojo should not handle byteranges at all, or handle them completely. but in the former case, is there any realistic alternative? i'm not sure if proxy servers can be configured to send ranges of upstream responses. probably not.
08:53 sri simple byteranges are commonly used to resume downloads
08:53 crab otoh, seeing multiple byte ranges in a request is extremely unusual. if it's implemented in core, it probably won't be exercised much.
08:54 sri browsers support simple byteranges, but not multipart
08:54 sri i don't agree with your all or nothing argument
08:55 sri since there is a lot of value in simple byteranges, but almost none in multipart byteranges
08:55 crab i don't agree with it either. that crab guy doesn't know what he's talking about.
08:55 sri (as far as we know)
08:55 sri heh
08:56 crab multiple byte ranges is (a) weird, (b) not used often, (c) not terribly hard to implement as a plugin, at least for static files.
08:57 sri if you have an opinion please express it in the github thread :)
08:57 crab i can do that?!
08:57 sri i insist on it
08:58 sri https://github.com/kraih/mojo/issues/184
09:06 sri ok, think i have expressed my position now
09:07 sri this might actually be really simple to do with a plugin
09:10 crab "Mostly used for custom dispatchers and postprocessing static file responses"
09:10 crab how would a custom dispatcher work? just do $whatever and set $self->res?
09:14 sri you could use ot for your own routes implementation
09:14 sri oh, that's what you said :)
09:15 crab yeah
09:15 sri $whatever and set $self->res is exactly how it works
09:16 * sri is still in pre-breakfast stage...
09:17 crab the only troublesome thing is, if the request has multiple byteranges, you'd need to disable mojo's handling of byteranges in before_dispatch somehow, otherwise the response will have only the first chunk in after_static_dispatch
09:17 sri nono
09:18 sri you can reuse that information
09:18 sri build a new repsonse based on it
09:18 sri saves you from dealing with files
09:18 sri you just build more byte range assets
09:18 sri and put them in a multipart container
09:18 crab my idea was, let the static dispatcher send the whole thing, then look at Ranges and substr() out the parts you need
09:19 crab but if mojo has already served up only the first requested range, then "the whole thing" would be too little.
09:19 * sri doesn't follow
09:20 sri nothing is served, you intercept the static response
09:20 crab give me a moment, let me look at the code.
09:21 * sri needs breakfast anyway
09:24 crab yes, i'm wrong, because of ->start_range/->end_range
09:24 crab i thought the static dispatcher would create a response with an asset that contained only the requested chunk
09:24 crab but the response has the full asset and just marks the range
09:25 crab so yes, i just look at that in after_static_dispatch and construct a new response.
09:25 crab easy peasy.
09:58 sri ye :)
09:58 sri xantus++ # for sane range support in core
10:31 sri hmm, those news reports from norway sound weird
10:31 sri 7 killed by bomb and 81! shot?
10:32 GabrielVieira2 joined #mojo
10:37 GabrielVieira left #mojo
10:56 crab terrible.
10:56 purl "Things come in three major degrees in the human experience, I think.  There's good, bad, and terrible.  And as you go down into progressive darkness toward terrible, it gets harder and harder to make subdivisions." -- Stephen King, _Rita Hayworth and Shawshank Redemption_
11:48 preflex_ joined #mojo
11:52 preflex left #mojo
12:05 sugar left #mojo
12:05 sri xantus[]: i could use the EV backend later today *poke*
12:06 sri ioloop rewrite is almost finished, just some TLS bugs left to fix
12:18 tabbi joined #mojo
12:18 tabbi left #mojo
12:25 marcus sri: looks like the bomb was a diversion. Pulling all the police to downtown so he'd have lots of time out on the island.
13:07 diego_k is now known as diegok
13:56 mattp_ left #mojo
13:57 mattp_ joined #mojo
14:16 crab that's an extremely sick and disturbing level of planning to kill a bunch of kids.
14:26 marcus crab: yepp. up to 85 confirmed dead on the island. Probably will be more that drowned.
14:38 * sri didn't know you had right wing crazies in norway
14:40 mspo is morbo supposed to pick up changes anywhere in your app?
14:43 mspo I'm trying to move my lite app into a regular app and I've lost the ability to use my helper.  I redefined it in the controller like $self->helper(foo => sub
14:43 mspo but now my template just gives an error about an undefined subroutine
14:48 crab redefined it in the controller? that sounds wrong.
14:48 crab helpers belong to the app, not the controller.
14:49 crab and: morbo picks up changes to lib and templates.
14:51 crab (but you can make it watch othr directories if you want)
15:02 mattp_ left #mojo
15:02 mattp joined #mojo
15:09 marcus http://www.dagbladet.no/2011/07/​23/nyheter/utoya/drap/17427253/
15:09 mspo oh, so in MyApp.pm?
15:17 mspo okay, got it working
15:18 mspo so I'm trying to rebuild my site with a few mojo apps.  Should I package them all up as embeddable plugins?
15:29 sri not unless you have a good reason to do so
15:31 sri http://mojolicio.us/perldo​c/Mojolicious/Plugin/Mount # this is a more general purpose solution
15:39 mspo what's the difference?
15:39 purl the difference is Epsilon, a number that gets closer to zero the longer you think about it.
15:46 GabrielVieira2 left #mojo
15:47 GabrielVieira2 joined #mojo
15:50 marcus mspo: with the mount plugin, your apps work standalone as well.
16:02 GabrielVieira2 left #mojo
16:04 trone_ is now known as trone
16:28 GabrielVieira2 joined #mojo
16:34 GabrielVieira2 left #mojo
16:59 sri i hate it when i just can't find that one last bug :S
17:00 crab it's under marcus's couch
17:03 sri naah, that's the evil roomba
17:10 crab it's inside the evil roomba.
17:16 vel joined #mojo
17:25 sh9 joined #mojo
17:28 ashleydev left #mojo
17:46 anaio joined #mojo
17:46 mattp_ joined #mojo
17:55 sugar joined #mojo
18:03 GabrielVieira2 joined #mojo
18:18 Kulag left #mojo
18:19 Kulag joined #mojo
18:31 sh9 left #mojo
18:35 Foxcool joined #mojo
19:09 GabrielVieira2 left #mojo
19:09 GabrielVieira joined #mojo
19:26 GabrielVieira left #mojo
19:26 GabrielVieira joined #mojo
19:45 sri i like the new safari a lot, but it is really missing some poweruser features
19:46 GabrielVieira left #mojo
19:46 GabrielVieira joined #mojo
19:51 anaio sri, i think the core tag helper url_for should have a nocache option, ... what do you think?
19:58 Drossel joined #mojo
19:58 sri anaio: that question needs more explanation
19:58 marcus_ joined #mojo
19:59 sugar_ joined #mojo
19:59 y0j_- joined #mojo
19:59 sri yay, more marcus
19:59 marc_ joined #mojo
19:59 mfridh_ joined #mojo
19:59 maluco_ joined #mojo
20:01 Kulag left #mojo
20:01 sugar left #mojo
20:01 mattp_ left #mojo
20:01 mfridh left #mojo
20:01 marcus left #mojo
20:01 maluco left #mojo
20:01 marc left #mojo
20:01 ajgb left #mojo
20:01 charsbar_ left #mojo
20:01 edenc left #mojo
20:01 y0j_ left #mojo
20:01 marcus_ is now known as marcus
20:02 marcus can never get enough marcus
20:02 sugar_ is now known as sugar
20:03 marcus the new xcode storyboard stuff is super neat.
20:06 crab what is it for?
20:06 mattp_ joined #mojo
20:06 ajgb joined #mojo
20:06 edenc joined #mojo
20:06 charsbar_ joined #mojo
20:08 xaka joined #mojo
20:14 marcus crab: It's an extension to interface builder to let you configure the flow between your scenes graphically
20:20 kaare left #mojo
20:30 marcus I'm reading the killer's book now. He's a religious fundie.
20:30 marcus freemason
20:38 abra left #mojo
20:49 abra joined #mojo
21:02 sri phew, last bug found
21:02 sri all tests pass again \o/
21:21 mattp left #mojo
21:21 mmp joined #mojo
21:33 GabrielVieira left #mojo
21:36 Foxcool left #mojo
21:38 perlite_ joined #mojo
21:39 sri oh shit
21:39 sri can someone please stress test the latest mojolicious from cpan on lion
21:40 sri "ab -c 30 -n 10000 -k http://127.0.0.1:3000/welcome" against a fresh generated app
21:41 sri i need memory usage and list of weird things happening
21:41 perlite left #mojo
21:41 perlite_ is now known as perlite
21:42 sri and perl version
21:42 sri i think IO::Poll is broken again
21:48 marcus sri: I think it sucks that mojo can't generate all lc app names.
21:49 marcus ( mojo generate app iwt )
21:49 sri marcus: feel free to improve
21:49 sri but don't forget to check for side effects
21:50 marcus sri: testing trunk now
21:51 sri that stuff has no tests, so it's very dangerous to mess with
21:52 sri main problem i see is that iwt doesn't survive camlize roundtrips
21:52 sri *+e
21:53 sri no idea if we depend on it somewhere (maybe another generator?)
21:53 alnewkirk|com left #mojo
21:53 alnewkirk|com joined #mojo
21:56 marcus sri: you mean because it would turn into Iwt ?
21:56 sri ye
21:57 sri that's the case you need to search for
21:58 sri if that's not a problem we migth be able to do it (hopefully no cpan module depends on it)
21:58 sri the only restriction is the generator though
21:59 sri even if it was risky you could always make your own
21:59 sri (maybe)
22:00 sri i'm just guessing… no idea if something depends on camelize roundtrips ;p
22:05 anaio sri, sometimes you want to trick a browser into not caching file by appending a random string to the url, ... i thought it would be cool to say <%= url_for '/foobar', nocache => 1 %>
22:06 sri anaio: just make your own helper
22:06 sri and while you are at it release it as a plugin ;)
22:07 sri <%= one_time_url_for '/lalala' %>
22:07 anaio i thought it would be a cool core feature
22:08 anaio you?
22:08 purl somebody said you was behind the times bro
22:08 sri well, it doesn't fit into url_for anyway
22:08 sri the syntax you suggested is already taken
22:09 sri it seems to be perfect for a plugin
22:10 anaio making it a plugin seems like overkill IMO
22:11 sri not at all
22:11 anaio what would be an appropriate name?
22:12 sri Mojolicious::Plugin::OneTimeURL?
22:12 anaio ... and the helper?
22:12 sri or actually OneTimeUrl
22:12 sri dunno, be creative :)
22:13 anaio k
22:16 marcus maybe you could make it support both random urls and incremental urls.
22:16 marcus I use that for static assets to bust the cache
22:16 marcus foobar.css?version=baz
22:16 Drossel left #mojo
22:16 marcus hmm, digest=sha would maybe be even better for that purpose.
22:18 mattp_ left #mojo
22:22 ashleydev joined #mojo
22:24 anaio marcus++
22:25 anaio marcus, ... pull the version from the conf? (if avail) or the top-level namespace?
22:25 Kulag joined #mojo
22:25 marcus anaio: pull it from the file you are linking to?
22:26 mmp left #mojo
22:27 mmp joined #mojo
22:33 sri i seriously need someone to run a stress test on os x
22:34 sri there is a small risk that the next release might be broken on os x
22:35 Kulag left #mojo
22:36 Kulag joined #mojo
22:37 anaio i notice lots of SaaS products giving account there own URL e.g. myname.webapp.com when it seems more pratical to use webapp.com/myname, ... any idea why (any advantages)
22:39 sri hmm… maybe sockets are broken under lion
22:39 anaio i mean ... from a routing perspective, seems easier
22:40 sri lots of broken pipe errors
22:40 sri wonder what changed
22:40 sri marcus: you heard anything about broken pipes in lion?
22:41 marcus sri: nope.
22:41 marcus sri: want me to try ab?
22:41 sri even a clean perl 5.14.1 install with only cpan mojolicious breaks
22:41 sri marcus: that would be nice
22:42 marcus sri: yes, I get broken pipes
22:42 sri maybe another xcode bug
22:42 sri it prolly breaks something in perl
22:43 marcus 28M private
22:43 sri yea
22:43 sri it goes up to 60M sometimes
22:43 sri and drops back to 30M
22:44 sri my bet is on llvm screwing with perl
22:44 marcus second run I got apr_socket_recv: Connection reset by peer (54)
22:44 sri ye seen that too
22:44 sri sockets got very fragile
22:45 Drossel joined #mojo
22:45 sri glad it's not just me :D
22:46 sri but worried about lion as a perl dev platform :/
22:46 Kulag left #mojo
22:46 sri wait a minute
22:47 * marcus waits a minute
22:47 sri marcus: have you tried something else than ab?
22:47 marcus sri: nope.
22:47 sri i mean… it's not like we die
22:47 marcus sri: my thought as well.
22:47 sri it could be ab too
22:47 marcus sri: I don't deploy to osx anyways..
22:49 sri broken pipe is a normal error, and memory usage could just be llvm being funny
22:50 sri now… googling for broken ab will be hard :S
22:51 sri HAHA!
22:51 sri "ab -c 30 -n 10000 -k http://mojolicio.us/" fails too
22:52 marcus gues it's ab. jolly good
22:53 * marcus tries installing autobench from brew
22:54 sugar left #mojo
22:56 sri you also have to patch EV to make it compile on lion
23:34 preflex joined #mojo
23:37 preflex_ left #mojo

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