Camelia, the Perl 6 bug

IRC log for #mojo, 2011-03-16

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

All times shown according to UTC.

Time Nick Message
00:04 ok22 left #mojo
00:11 ok22 joined #mojo
00:13 Akron Good night everyone.
00:13 Akron left #mojo
00:18 ok22_ joined #mojo
00:23 ok22 left #mojo
00:41 perlrocks Twitter: "Mojo:: Client has been deprecated-Mojolicious | Google Groups http://bit.ly/i8wiiW" (ja) --wPerlnews http://twitter.com/wPerlnew​s/status/47819795881660416
00:55 xris left #mojo
00:59 cxreg joined #mojo
01:24 daviddelikat1 joined #mojo
01:28 daviddelikat left #mojo
01:48 perlrocks Twitter: "135x17 Scarilyn print by mojolicious on @Etsy http://etsy.me/hmS3Xo FINALLY LAUNCHED THE #etsy store... lots more to come soon!!! xoxoxoxo" --mojo_licious__ http://twitter.com/mojo_licio​us__/status/47836493485572096
02:20 trcjr joined #mojo
02:28 DaTa left #mojo
02:29 alnewkirk joined #mojo
02:46 ZadYree left #mojo
02:59 DaTa joined #mojo
03:04 marcio_ferreira left #mojo
03:07 alnewkirk left #mojo
03:13 crab didn't there used to be a "label" helper?
03:53 perlrocks Twitter: "Awesome RT @TheWellingtonSD: We are mojolicious! How about you? RT @MomWithMojoSD: How is everyone's Mojo today?" --MomWithMojoSD http://twitter.com/MomWithMoj​oSD/status/47868152872312832
04:07 DaTa left #mojo
04:22 ispy_ joined #mojo
04:22 Foxcool joined #mojo
04:29 grim_fandango left #mojo
04:32 Foxcool left #mojo
04:38 DaTa joined #mojo
04:41 ispy_ left #mojo
04:46 DaTa left #mojo
04:51 ispy_ joined #mojo
04:51 ok22_ left #mojo
05:02 ok22_ joined #mojo
05:08 DaTa joined #mojo
05:17 ispy_ left #mojo
06:07 koban` joined #mojo
06:11 giwi joined #mojo
06:13 kaare joined #mojo
06:15 DaTa left #mojo
06:24 ysyrota joined #mojo
06:46 DaTa joined #mojo
06:46 koban` left #mojo
06:53 ok22_ left #mojo
06:56 koban joined #mojo
06:57 ver joined #mojo
07:04 ok22_ joined #mojo
07:09 esskar joined #mojo
07:14 dekimsey left #mojo
07:15 dekimsey joined #mojo
07:26 cosmincx joined #mojo
07:27 nuba joined #mojo
07:34 Christian joined #mojo
07:34 Christian morning
07:43 marcus sri: ping?
07:45 cosmincx left #mojo
07:50 crab good morning christian, marcus.
07:52 marcus sri: doesn't the async dns resolver in mojo cache at all?
07:52 crab it doesn't afaict
07:52 marcus :/
07:52 marcus can it be disabled?
07:53 stephan48 21:55:57 -!- stephan48 was kicked from #HvK by Bandit [Bandit]
07:53 stephan48 sry
07:53 fhelmber_ joined #mojo
07:53 crab can what be disabled?
07:54 marcus crab: the async resolver?
07:54 crab the resolver itself? there's no mechanism to disable it
07:54 marcus ok
07:54 crab it's really just an IOLoop method that issues the right sort of DNS query
07:55 marcus we had a situation at $client yesterday where one of the dns servers in resolv.conf wasnt responding
07:55 marcus and there was much blocking...
07:56 crab (mojo's resolver doesn't know about resolv.conf, does it? i thought you had to set a server explicitly)
07:56 crab (aside, because obviously even if you set it explicitly, the same problem would happen if it stopped responding)
07:57 marcus crab: humm, so it's not enabled by default then?
07:57 marcus (mojo::client doesn't use it?)
07:57 crab marcus: it defaults to using 8.8.8.8
07:57 crab and i'm wrong, it now seems to read resolv.conf too
07:58 koban left #mojo
07:58 crab and it uses the last one.
07:59 marcus yeah
08:00 marcus that's pretty weird
08:00 marcus it should probably default to using the *first*?
08:00 koban joined #mojo
08:01 crab that's the generally expected behaviour
08:01 crab and to fallback to the next one if it fails
08:02 marcus yeah. This seems a bit fragile
08:04 cosmincx joined #mojo
08:09 breaker313 joined #mojo
08:14 koban left #mojo
08:14 koban joined #mojo
08:14 marcus it should probably read all the resolv.conf nameservers, and skip to the next one on error/timeout
08:15 marcus would also be nice if it respected options timeout
08:16 marcus and maybe options attempts
08:16 crab hm. should be doable.
08:17 crab i'll have a crack at it on the weekend or so.
08:17 koban left #mojo
08:17 koban joined #mojo
08:24 marcus crab++
08:25 minimalist left #mojo
08:33 Foxcool joined #mojo
08:46 koban left #mojo
08:46 koban joined #mojo
09:35 Foxcool left #mojo
09:48 baton8 left #mojo
09:52 Foxcool joined #mojo
10:02 Foxcool left #mojo
10:20 spleenjack joined #mojo
10:23 sri marcus/crab: it is absolutely possible, the problem is testing
10:23 marcus sri: The current implementation is much worse than just using libc
10:23 sri marcus: bullshit
10:23 marcus sri: For a serious business, depending on one dns server is hopeless
10:23 marcus it's a huge risk
10:23 marcus single point of failure
10:23 purl single point of failure is that there's one instance of an operating system instead of two
10:24 sri use google dns then
10:24 marcus that's still a single point of failure.
10:24 marcus that's why google provides two
10:27 sri that's not true, google dns is not a single point of failure
10:27 marcus (8.8.8.8 and 8.8.4.4)
10:27 sri each ip leads to a huge cluster
10:27 marcus sri: you can say what you want, the sysadmins of big companies won't accept that.
10:28 sri so?
10:28 marcus so, it's much worse
10:28 sri blocking is worse than non-blocking
10:28 crab 8.8.8.8 is anycasted to... yeah, huge clusters.
10:28 marcus sri: downtime is worse than no downtime
10:29 sri marcus: that is your fault
10:29 sri if your dns sucks your dns sucks
10:29 sri anyway, feel free to improve support for bad dns servers
10:29 crab sometimes it's hard to avoid multiple inheritance.
10:30 crab i wish i could tell a class "your base class is actually this augmented derivative of what you thought was your base class"
10:33 sri marcus: btw. 8.8.8.8 and 8.8.4.4 leads to the same datacenter from here, quite sure they are just aliases for the exact same cluster
10:35 crab it's not one physical cluster, it choose the dns server(s) closest to you using anycast
10:35 sri yes
10:36 crab but the machines that respond are distributed across the world
10:39 sri i really wouldn't mind another safety net, but i doubt someone would be willing to write the tests for it
10:41 crab hard to test the case of what happens with an unreachable server robustly, agreed
10:42 GitHub4 joined #mojo
10:42 GitHub4 mojo: master Sebastian Riedel * b9c6487 (1 files in 1 dirs): fixed typo - http://bit.ly/geQVl4
10:42 GitHub4 left #mojo
10:43 sri exactly, and a safety net without reliable tests is not worth anything
10:50 crab guess i won't be working on it this weekend then
10:50 sri :(
10:50 sri it all depends on marcus then
10:51 marcus yeah, marcus is switching to a reliable client instead.
10:51 marcus :p
10:51 sri you sure being rude is constructive?
10:51 crab well, i wouldn't mind writing the code if someone can suggest a way to test that will get the patch accepted
10:52 sri you know, before you started acting like a jerk i was considering helping
10:52 sri now i'm going to do something fun
10:52 crab me?
10:52 purl you are, like, dreadfully tempted to filter-branch your master to rewrite some commits where your address is wrong (because of the hardcoded map for ams in git-p4raw :), but i've been resisting so far. or a famous detective and related to dr zoidberg
10:52 sri marcus
10:52 marcus sri: are you saying you're not rude?
10:54 sri marcus: now i am yes
10:55 marcus sri: you sure know how to alienate your community. No wonder people get burned out.
10:55 sri wtf?
10:55 sri what are you expecting?
10:56 marcus pretty much this, I guess?
10:56 sri i mean when you started such an unconstructive discussion
10:56 marcus you started by telling me my complaints were bullshit
10:57 sri my first sentence was showing you the problems
10:57 sri you decided to ignore that
10:57 sri and just go on about how shitty the resolver is
10:57 sri address the issue and improve it
10:59 sri you can't expect more than me accepting patches, this is open source
11:00 sri [11:23] sri: marcus/crab: it is absolutely possible, the problem is testing
11:01 sri i was trying to be constructive there, you dragged the discussion down
11:02 * crab sighs
11:02 sri marcus: you countered that with "The current implementation is much worse than just using libc" to which i said "bullshit"
11:03 sri because i was expecting a constructive discussion about testing the issue
11:05 marcus sri: well, I was already talking about fixing it before you appeared, but your attitude made it sound like there's no problem, and it's just us mortals being stupid.
11:05 marcus very elitist attitude.
11:06 marcus anyways, I'll try to look into writing something to address the issue.
11:06 sri i was just reacting to your claims, out of context
11:06 marcus and parsing them in the worst way possible
11:06 marcus I think you could irritate a rock, on a good day.
11:07 crab i am a rock, i am an iiiisland
11:07 sri the alternative is me going back to simply asking for a broken test case
11:08 sri maybe i should reduce community interaction a bit
11:08 sri (for me)
11:08 marcus sri: also, you should realize that the way we're using mojo::client today (in a fastcgi catalyst app), async doesn't really matter to us.
11:09 sri marcus: what's the point of that sentence?
11:09 marcus it might be a huge win in the future tho.
11:09 marcus sri: Just saying, in some cases it might be better to use the libc resolver.
11:10 marcus which also caches requests
11:10 sri you're ignoring all the issues here
11:10 sri if you think there is a better way to do something, please try it and if it really is better provide a patch
11:11 marcus I'm comparing using mojo to using lwp, for instance.
11:11 sri which makes no sense
11:11 sri try unit testing mojolicious apps with lwp for example
11:11 sri those are the issues you're ignoring
11:12 marcus but we're not using it in a mojolicious app
11:12 marcus are you saying mojo::useragent is *only* for mojolicious web apps?
11:12 jwang left #mojo
11:12 sri that's not at all what i'm saying
11:12 sri it is *one* of the use cases
11:12 marcus I was just suggesting it might be useful to be able to disable the async resolver in some cases.
11:12 marcus if you agree it might be useful, I might come up with a patch ioloop to do that.
11:12 sri and a huge testing problem
11:14 marcus sri: option to disable async would be a huge testing problem?
11:14 marcus async resolving even
11:15 sri you would need to make sure it works in all use cases
11:16 sri ok wait a minute
11:16 sri have you actually looked at how the resolver works now?
11:17 sri it will attempt a non-blocking dns lookup and if that doesn't work (error or timeout) fallback to blocking libc lookup
11:18 marcus so I could just set the timeout to 0 seconds? ;)
11:18 marcus sri: we noticed the fallback behaviour, which made our app *very slow* when the last dns server in resolv.conf was down.
11:20 sri that is correct, you could set dns timeout to 0
11:21 marcus k. guess that would lead to some unused code, but it would work I guess.
11:22 sri if you go blocking that's the least of your problems
11:24 marcus sri: but since we're doing a blocking get, we're already blocking?
11:25 sri libc lookups usually take longer
11:26 marcus not if it's cached tho?
11:26 GitHub50 joined #mojo
11:26 GitHub50 mojo: master Sebastian Riedel * eaea44e (2 files in 2 dirs): changed resolv.conf parser in Mojo::IOLoop to use the first nameserver - http://bit.ly/fKGyZJ
11:26 GitHub50 left #mojo
11:26 marcus we keep hitting the same server over and over :)
11:26 sri cached is not very fast either
11:29 crab it's too bad there's no easy way to know if there's a recursive resolver running on 127.0.0.1:53, otherwise i'd default to that or use 8.8.8.8 or let the user override via %ENV, and just not touch resolv.conf at all.
11:29 crab because touching resolv.conf means people will expect you to behave the way libc's resolver behaves, which means duplicating various behaviours.
11:30 crab if i were running an app which wanted fast DNS responses, i'd be running bind or something on localhost with forwarders and querying it, and it would take care of retrying and falling back to the next forwarder and all that.
11:33 sri switching between alternative dns servers would be a sensible solution
11:33 sri if someone solves the testing problem
11:34 sri we also have Mojo::Cache now, in case a simple in memory lookup cache makes sense
11:34 marcus was thinking that too
11:34 crab if you or marcus (or anyone) has any ideas about that, i am still happy to write the code
11:35 crab i don't think mojo should be caching nameserver responses
11:35 marcus sri: might be a problem that cache expiry only is by number of requests, not time
11:35 crab either you do it right, which means you care about ttl and expiry and all of that stuff, or you do it wrong and break someone's expectations at an inconvenient moment
11:35 sri there are many problems
11:35 marcus if we cache nameserver responses, we should respect zone rules
11:35 marcus yeah, it needs to respect ttl
11:37 crab the current resolver more or less assumes that it'll be talking to a forwarder/resolver, it's not a bad assumption to think that server will cache for you. and ifyou want the cache closer, you run the forwarder closer.
11:38 crab i strongly feel that having lots of little ns caches is counterproductive
11:38 marcus crab: libc and windows caches dns responses locally tho
11:38 marcus not doing a network request is quite a bit faster than doing a network request
11:39 sri udp is pretty damn fast
11:39 crab marcus: that's the system resolver. it makes some sense for that to cache. but you wouldn't expect individual apps talking to the system resolver to ALSO cache the responses
11:39 sri most latency is caused by tcp overhead
11:40 marcus crab: mojo doesn't talk to the system resolver tho.
11:40 crab in this case, we're not using the system resolver because it's blocking
11:40 crab right
11:40 crab marcus: is there any reason you can't run a bind on localhost, set its forwarders {} to whatever your upstream dns servers are, and make mojo query 127.0.0.1?
11:40 marcus sri: I thought dns used udp? :)
11:40 crab dns uses both udp and tcp
11:40 sri exactly, it does
11:41 sri tcp only as a fallback though
11:41 crab (i say bind, but any of the other lightweight but complete resolver implementations would do)
11:42 crab if you truly can't afford the cost of exchanging two udp packets with localhost, then it wouldn't work, but aren't you just going to make an out-of-network request to a web server afterwards anyway?
11:42 sri what you also shouldn't forget is that your app actually kept working even though your main dns server went down
11:42 sri slower, but working
11:42 sri and so far we've had not a single report of dns trouble
11:43 crab long timeouts because of dead dns servers is pretty much a traditional problem
11:43 sri so maybe baby steps would be smart
11:43 sri like server alternatives
11:43 marcus sri: actually, not my main dns server, the last dns server :)
11:43 sri marcus: well, that issue has been addressed already
11:43 marcus yeah I noticed
11:43 marcus sri++
11:45 sri GROUPHUG
11:45 purl couldn't get the headlines: http://grouphug.us/xml wasn't successful
11:47 sri switching servers once you didn't get a response from the non-blocking resolver is quite trivial
11:47 marcus purl: literal GROUPHUG
11:47 purl marcus: GROUPHUG =is= <rss = "http://grouphug.us/xml">
11:48 marcus purl: no, GROUPHUG =is= <rss = "http://feeds.feedburner.com/grouphug/confessions">
11:48 purl marcus: excuse me?
11:48 marcus purl: no, GROUPHUG is <rss = "http://feeds.feedburner.com/grouphug/confessions">
11:48 purl okay, marcus.
11:48 marcus GROUPHUG
11:48 purl marcus: 56812166; 531652406; 980328324; 917721091; 593628956; 599026524; 830752497; 502225403; 206056690; 146427707
11:48 marcus oj
11:49 sri purl: SRIBLOG is <rss = "http://feeds.feedburner.com/krai">
11:49 purl OK, sri.
11:49 sri purl: sriblog?
11:49 purl couldn't get the headlines: http://feeds.feedburner.com/krai wasn't successful
11:49 sri eep
11:49 sri purl: no, SRIBLOG is <rss = "http://feeds.feedburner.com/kraih">
11:49 purl okay, sri.
11:49 sri purl: sriblog?
11:49 purl sri: Mojolicious 1.12: IPv6 goodness; A logo for Perl 5; Interview about Mojolicious; Mojolicious 1.1: Awesome features on a caturday; Mojolicious and Plack; Mojolicious is changing the game; The truth; What does a normal Mojolicious application look like?; Mojolicious 1.01: TLS goodness; Mojolicious 1.0 released; Actual Perl Marketing; Perl is ready for HTML5; A logo for Perl; Glue for the web; Mojolicious is awesome
11:49 sri interesting
11:50 sri anyway, back to dns
11:50 sri the one big problem that needs to be solved is testing alternative servers
11:51 sri maybe with a 0 timeout
11:51 sri that could work
11:51 sri you make one lookup with timeout 0 and the next with 3 again
11:52 crab that doesn't really test it truthfully though
11:52 sri or we could use a host that would never run a dns server
11:52 sri that's all we need for a live test
11:53 marcus that's easy
11:53 marcus 192.0.2.0/24
11:53 marcus Assigned as "TEST-NET" for use solely in documentation and example source code and should not be used publicly.
11:53 sri that could work
11:53 * marcus & # squash
11:53 sri crab: seems you have a test scenario
11:54 sri should be fine in t/mojo/ioloop_online.t
11:55 sri Mojo::IOLoop->dns_server is also experimental
11:55 sri you can replace it with dns_servers
12:04 mambo joined #mojo
12:06 mambo Hello, what following Mojo error could mean ? Can't use string ("Projector::Controller::Project") as a HASH ref while "strict refs" at .../Mojolicious/Controller.pm line 719
12:07 lee7 left #mojo
12:07 mambo never mind, found the problem
12:07 crab it's a perl error, not a mojo error. look in perldiag.
12:15 Foxcool joined #mojo
12:24 marcio_ferreira joined #mojo
13:08 kaare left #mojo
13:10 GitHub188 joined #mojo
13:10 GitHub188 mojo: master Sebastian Riedel * f9fe372 (3 files in 3 dirs): added support for multiple dns servers to Mojo::IOLoop - http://bit.ly/fGDvKy
13:10 GitHub188 left #mojo
13:10 sri once there's a test case it is all trivial
13:12 * sri pokes marcus and crab
13:30 fmerges joined #mojo
13:40 dotan joined #mojo
13:43 crab ...oh
13:44 crab you did it already?
13:44 crab oh well.
13:52 crab looks good.
13:55 sri gonna clean it up a bit though
14:00 kaare joined #mojo
14:02 giwi left #mojo
14:03 marcus hai
14:03 marcus sri++
14:04 marcus sri: you respond well to abuse ;)
14:04 sri i feared that impression :(
14:04 stephan48 you abused sri? how could you!
14:05 marcus stephan48: won't someone think of the children? :/
14:05 sri i'm seriously considering to delay the commit by a week
14:05 marcus sri: isn't it too late for that?
14:05 stephan48 marcus: no
14:05 sri current commit is broken
14:06 sri so i still havethe option
14:06 marcus you probably won't ship for a week or two either way.
14:06 marcus I'm just glad it's coming =)
14:09 sri i suppose it's the only sensible thing to do
14:12 sri marcus: actually i want to make more regular small releases now
14:13 sri so it could have been out tomorrow
14:15 crab delay by a week? why?
14:15 sri it would send the wrong signal
14:15 sri "if you behave bad enough sri will implement it for you"
14:16 ComLock joined #mojo
14:16 crab oh
14:17 crab "if you behave bad enough sri will implement it for you a week later" ;-)
14:17 sri suggestions welcome :)
14:17 marcus it's not like I care :)
14:18 marcus this week, next week, whatever
14:25 marcus while we're talking about dns - http://www.pinkbike.com/news/DN​S-Prefetching-implications.html
14:25 marcus sri: you're awesome -.-
14:27 crab sri: i'm sorry, but i think you're overreacting
14:27 crab marcus: clearly their bike should have been blue
14:28 marcus crab: don't worry, he's just working on his marketing again ;-)
14:29 sri crab: zero tolerance for rude behavior here, i made a mistake earlier joining in and i don't intend to repeat it
14:29 crab well, i don't think marcus was trying to be rude, but ok.
14:30 sri "sri: you respond well to abuse ;)"
14:30 sri if that isn't then i'm very confused now
14:30 marcus yeah, I think you're pretty confused if you thought that was rude
14:30 crab it was a joke
14:30 marcus it was
14:31 marcus anyways, I'm heading out to pick up $daughter at kindergarden and drive crates.
14:31 marcus sorry if you thought I was trying to insult you, sri. really wasn't
14:31 crab drive... crates?
14:32 sri ok, then i'm clearly not capable of this english thing
14:32 marcus crab: I'm moving.
14:32 marcus to a new apartment
14:32 crab oh. i had this vivid mental image of a crate with a steering wheel...
14:33 marcus :D
14:33 marcus that would be awesome
14:33 marcus actually, it's just carton boxes
14:34 marcus not really crates at all.
14:34 * sri is starting to understand the japanese perl hackers
14:35 crab sri: oh no, quick, dip your head into bleach!
14:35 * sri dips head into bleach
14:35 stephan48 whats bleach?
14:35 purl rumour has it bleach is a dilute sodium hypochlorite (NaOCl) solution or the first Nirvana album. or tasty
14:36 sri mmm...bleach
14:36 stephan48 bleiche?
14:36 crab stephan: Bleiche
14:36 crab yes
14:39 GitHub42 joined #mojo
14:39 GitHub42 mojo: master Sebastian Riedel * ec0a7d1 (7 files in 4 dirs): cleanup - http://bit.ly/eq4EGn
14:39 GitHub42 left #mojo
14:39 sri maybe i just need to be more apathetic
14:42 * tempire puts on his sheriff hat
14:42 sri finally!
14:42 sri i'm starting to get the impression you're always hiding
14:43 tempire away from the computer a lot, playing politics
14:43 sri excuses excuses ;p
14:43 stephan48 a politican?
14:44 tempire indeed, it seems so
14:44 tempire it's been a week since I wrote any code.
14:44 * tempire slips down the rabbit hole
14:44 koban left #mojo
14:45 marty WOW!  I just came up for air (this is the busiest time of year for business and I have no time for much programming) and I see all these really KEWL new things in Mojolicous!  Wow
14:45 marty sri+++++
14:45 tempire sri: the fact that you respond and do something about peoples complaints shows that you're interested in solutions.
14:45 tempire but you come across as a jerk initially.
14:45 sri i wonder why :/
14:45 tempire it seems to be a matter of whether the person sticks around long enough to see the reality
14:46 tempire marcus++
14:46 marty I love the routing shortcuts thing.   I am sure I can use that...
14:47 tempire It's fine for irc veterans, 'cause you develop a thick skin.  especially if you hang out in any channels with mst & crew.
14:47 Foxcool left #mojo
14:48 sri i usually don't have the problem in german
14:48 tempire sri: my only recommendation is to steer clear of any negative words.
14:49 tempire ex: sucks, bullshit, etc
14:49 sri that reduces my vocabulary significanty :(
14:49 tempire bullshit
14:49 tempire ;)
14:49 marty Crappadoodle sri!!!
14:50 * sri adds that to his vocabulary
14:52 tempire I knew you were a bit
14:52 tempire bot, even
14:53 tempire sri, bullshit is crappadoodle
14:53 tempire sri what is bullshit?
14:54 sri rumor has it bullshit is crappadoodle
14:54 GitHub190 joined #mojo
14:54 GitHub190 mojo: master Sebastian Riedel * 2df01f9 (4 files in 3 dirs): more cleanup - http://bit.ly/hZ5pH9
14:54 GitHub190 left #mojo
14:54 crab yes, that's it. reform your personality to match that of purl! everyone loves purl! (well, except me)
14:55 sri that sounds like the logical conclusion
14:56 tempire purl, crab is full of rainbows for purl
14:56 purl ...but crab is a famous detective...
14:57 sri :D
14:57 ver left #mojo
15:00 sri maybe i need a few standard phrases
15:02 ComLock left #mojo
15:02 sri like those call center bots
15:02 spleenjack left #mojo
15:03 sri multiple dns servers should be fine now btw
15:04 sri if a lookup fails it will fallback to blocking and the next one will try a new dns server... and so on
15:05 sri so you'll have a few slow requests until it finds a new working dns server
15:05 sri none will fail though
15:05 stephan48 if it cant find working dns server?
15:05 sri then requests say slow
15:05 sri *stay
15:05 sri until one dns server comes back
15:06 stephan48 k
15:06 stephan48 what will it use when all dns fail?
15:06 sri the two google ones are the last in line though, i doubt they'll ever be offline
15:06 stephan48 k
15:06 sri libc resolver
15:07 sri if that fails your requests will fail
15:07 stephan48 k
15:07 spleenjack joined #mojo
15:08 sri btw. i don't think dns caching will be relevant for useragent
15:08 sri since you have keep alive anyway
15:10 sri which has no limit besides a soft 15 seconds inactivity timeout
15:11 sri if you're losing that connection i would want a new dns lookup
15:11 sri to check if servers changed
15:15 ispy_ joined #mojo
15:17 mambo left #mojo
15:26 giwi joined #mojo
15:31 Kulag left #mojo
15:31 Kulag joined #mojo
15:35 Kulag left #mojo
15:37 Kulag joined #mojo
15:40 perlrocks Twitter: "#perl and #mojolicious make me like my job again...." --ericrost http://twitter.com/ericros​t/status/48046000316755968
15:44 Kulag left #mojo
15:46 giwi left #mojo
15:48 Kulag joined #mojo
15:54 Kulag left #mojo
15:55 Kulag joined #mojo
16:03 Kulag left #mojo
16:14 Christian nice evening
16:15 Christian left #mojo
16:17 Kulag joined #mojo
16:18 sri btw. it's quite fun to activate MOJO_IOLOOP_DEBUG=1
16:19 sri MOJO_IOLOOP_DEBUG=1 TEST_ONLINE=1 ./script/mojo test t/mojo/ioloop_online.t
16:24 giwi joined #mojo
16:25 Kulag left #mojo
16:34 cosmincx left #mojo
16:40 Kulag joined #mojo
16:43 ok22_ left #mojo
16:46 ok22_ joined #mojo
16:46 ysyrota left #mojo
16:54 breaker313 left #mojo
17:14 dotan left #mojo
17:18 ispy_ left #mojo
17:24 GitHub62 joined #mojo
17:24 GitHub62 mojo: master Sebastian Riedel * 58a3d52 (2 files in 2 dirs): fixed warning - http://bit.ly/hvMrHF
17:24 GitHub62 left #mojo
17:27 xris joined #mojo
17:32 sri DaTa: there were complaints about yamlconfig requiring perl 5.10 on the mailing list
17:38 perlrocks Twitter: "@vtivti are you still maintaining the Mojolicious plugins you wrote? If not, I'd like to take over Mojolicious::Plugin::Session" --perlninja http://twitter.com/perlninj​a/status/48075674082217984
17:49 minimalist joined #mojo
17:58 fmerges left #mojo
18:06 sri http://kraih.spreadshirt.com # raptor shirts are back in the us store!
18:07 sri took like forever to get the images reapproved :(
18:14 DaTa sri: *g* i didn't reckon anybody is using it beside me :)
18:14 sri i actually heard from a few people using it ;)
18:16 sri you only really get to know your users when you break something :D
18:26 fhelmber_ left #mojo
18:44 spleenjack left #mojo
19:06 giwi left #mojo
19:16 lee7 joined #mojo
19:23 tempire sri: I'm using this in a helper: $self->client->async->ioloop->handle( $self->tx->connection );
19:23 sri deprecated
19:23 tempire right
19:24 sri Mojo::IOLoop->singleton->handle(...)
19:24 tempire k
19:25 sri you can now rely on the singleton, it will always be there and running
19:25 sri (except of course for more restrictive environments like cgi, psgi...)
19:29 sri oh, mongodb 1.8 is out
19:36 perlrocks Twitter: "Mojolicious-Plugin-SslAuth 0.03 by TEMPIRE - http://frepan.org/~tempire/Moj​olicious-Plugin-SslAuth-0.03/" --cpan_new http://twitter.com/cpan_ne​w/status/48105424154669056
19:39 abra joined #mojo
20:31 perlrocks Twitter: "@perlninja It's pretty much obsolete since embedded mojolicious sessions." --vtivti http://twitter.com/vtivti/status/48119306915028992
20:32 Akron joined #mojo
20:44 sri actually there are still valid use cases
21:10 marcio_ferreira left #mojo
22:20 kaare left #mojo
22:23 GitHub79 joined #mojo
22:23 GitHub79 mojo: master Sebastian Riedel * f46df11 (3 files in 3 dirs): fixed small autoclosing tag bug in Mojo::DOM - http://bit.ly/fKf5Ub
22:23 GitHub79 left #mojo
22:25 sri there has to be a cleaner strategy to handle autoclosing tags...
22:37 Akron left #mojo
22:48 perlrocks Twitter: "Mojolicious-Plugin-YamlConfig 0.1.1 by DATA - http://frepan.org/~data/Mojoli​cious-Plugin-YamlConfig-0.1.1/" --cpan_new http://twitter.com/cpan_ne​w/status/48153770969931776
22:49 DaTa bye bye smartmatch
22:53 stephan48 left #mojo
22:53 stephen left #mojo
23:06 stephen joined #mojo
23:20 GitHub132 joined #mojo
23:20 GitHub132 mojo: master Sebastian Riedel * 7566517 (3 files in 3 dirs): fixed another optional tag bug - http://bit.ly/gc83Hr
23:20 GitHub132 left #mojo
23:24 stephen1 joined #mojo
23:30 stephen left #mojo

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