The web in a box - a next generation web framework for the Perl programming language

IRC log for #mojo, 2016-01-07

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

All times shown according to UTC.

Time Nick Message
00:07 miller joined #mojo
00:37 PryMar56 joined #mojo
01:00 jontaylor joined #mojo
01:07 disputin joined #mojo
01:14 disputin joined #mojo
01:18 PryMar56 joined #mojo
01:23 kaare joined #mojo
01:40 jberger sri: thinking about zero downtime restarts, I can change secrets via config on restart, correct?
01:40 jberger Otherwise the perfect forward stuff isn't as useful
01:40 sri of course you can
01:40 jberger I've just never done it
01:41 jberger So it's only hypnotoad options that can't change
01:41 sri you can't change the pid file location, that's about it
01:41 jberger Ok
01:41 jberger Good to know
01:41 sri most hypnotoad options can change
01:42 sri (i think all others, but i'm not 100% sure)
01:42 sri changing the listen locations is also not recommended i guess
01:43 sri since the old descriptors are kept
01:44 sri so, it works, but you'll leak descriptors every time you do it
02:03 binlei joined #mojo
02:07 * sri wonders if items on the todo list should get a deadline
02:08 sri i mean, none of these are must have features https://github.com/kraih/mojo/labels/future
02:09 sri and if there's not enough interest to get it done in like a year...
02:10 sri unix domain socket support has been unofficially on the todo list forever, maybe at some point it's time to say no
02:11 sri signatures has had 6 months and we are not even close to a roadmap
02:17 mattp sri: are you taking a break from active dev?
02:18 sri yea, have for the past two weeks already
02:23 sri so, http/2 and deps are long term goals and can stay i guess
02:24 sri server-side sni is kinda relevant there, and actually super simple to implement
02:24 sri guess that makes it a "maybe"
02:25 sri https://github.com/kraih/mojo/issues/830
02:25 sri https://github.com/kraih/mojo/issues/883
02:25 sri those are definitive candidates for failure
02:26 sri https://github.com/kraih/mojo/issues/882
02:26 sri that's the maybe
02:27 sri https://github.com/kraih/mojo/issues/891
02:27 sri and i guess this will die because of the database requirement
03:04 noganex joined #mojo
03:40 jberger preaction's Mercury project is an interesting way to possibly mitigate the db requirements
04:17 bpmedley https://bitbucket.org/snippets/bpmedley/dnLba <-- How about something like that?  Allowing for configuration options to be passed to Mojo::Base?
04:17 zivester joined #mojo
04:44 inokenty joined #mojo
04:52 sri i think #882 is very easy now https://github.com/kraih/mojo/issues/882#issuecomment-169543994
04:52 sri if nobody else wants to give it a try i might do it in the next few days
04:52 sri would be perfect for someone getting started with mojo contributions though
04:54 sri like 4 lines in Mojo::Server::Daemon::_listen to turn example.com_cert=server.crt into {'example.com' => 'server.crt'}...
04:54 sri a custom cert in t/mojo/certs/ for a random host like example.com
04:55 sri and some copypasta from t/mojo/websocket_proxy_tls.t into t/mojo/daemon_ipv6_tls.t with a fake CONNECT proxy
04:56 sri fake proxy only connects to your one test daemon, then you don't have to deal with anything dns
04:56 sri maybe an hour of work
04:57 sri or maybe one of our inactive core devs would like to commit something again ;p
06:09 bpmedley sri : https://github.com/kraih/mojo/compare/master...brianmed:issue_882 <-- How is this and how can I improve the test?
07:02 dod joined #mojo
07:07 dod joined #mojo
07:08 janus joined #mojo
07:13 trone joined #mojo
07:45 trone joined #mojo
08:21 Vandal joined #mojo
08:23 henq joined #mojo
08:23 AndrewIsh joined #mojo
08:41 McA joined #mojo
08:52 sugar__ joined #mojo
09:00 sh4 joined #mojo
09:07 cpan_mojo Mandel-0.29 by JHTHORSEN https://metacpan.org/release/JHTHORSEN/Mandel-0.29
09:48 kmx joined #mojo
09:55 bpmedley_ joined #mojo
10:18 jontaylor joined #mojo
11:00 bpmedley joined #mojo
11:09 kes joined #mojo
11:16 osfabibisi joined #mojo
11:16 osfabibisi bah, sorry if anyone replied to my question about Mojo::IOLoop above, I seem to have fallen off irc at some point
11:20 pink_mist I can't seem to find your original question in the first place .. mind repeating it?
11:27 osfabibisi perhaps I was already disconnected... hahahaha, sorry
11:27 osfabibisi we have a "log replayer" written with Mojo::IOLoop::Stream
11:28 osfabibisi e.g. it reads logs from one machine, and replays the same requests to another one (for load testing and cache warming etc.)
11:28 osfabibisi we sometimes find that the target machine will start panicking at around 50% load (even though it can obviously handle 100% load in a live situation)
11:28 osfabibisi and one of the fingers of suspicion is that the Perl script itself can't handle the task.  e.g. that it's being slow to receive the response from the server, and that is then causing the server to max out its threads waiting for the client
11:28 osfabibisi so... the question is: any immediate suggestions for things to investigate or optimize on the Mojo::IOLoop side of things?
11:29 jontaylor joined #mojo
11:29 pink_mist the only thing I could suggest is trying a different Mojo::Reactor backend for the IOLoop
11:30 osfabibisi ok, I can try that.  any recommendation?
11:31 pink_mist well, I think the default is Mojo::Reactor::Poll unless you have EV ... maybe try installing EV?
11:32 pink_mist there's also Mojo::Reactor::Epoll you could install from cpan
11:36 bpmedley joined #mojo
11:37 bpmedley osfabibisi : Have you profiled the code?
11:43 osfabibisi thanks pink_mist
11:43 osfabibisi not yet bpmedley
11:43 osfabibisi I'll try that before / after changing the backend
11:44 pink_mist there's also the question if you've written your webapp async'ly
11:44 osfabibisi the webapp is MarkLogic (an XQuery database server)
11:45 osfabibisi it claims to have 16 threads * 6 hosts
11:45 osfabibisi the problem with spiking is that each of those hosts maxes on threads basically
11:45 pink_mist threading and async are somewhat orthogonal
11:45 pink_mist you don't need threads to do things async'ly
11:46 osfabibisi sure
11:46 pink_mist oh, in fact, I think the EV backend won't even support threading
11:46 osfabibisi (I don't even know if what MarkLogic calls "threads" are in fact threads as such.)
11:47 pink_mist (perl's threads are less than ideal, and officially discouraged)
11:47 osfabibisi we just make HTTP calls to it
11:47 osfabibisi sure, we don't need to use threads in the perl client
11:47 pink_mist ok, good =)
12:14 denny joined #mojo
12:19 asarch joined #mojo
12:20 denny How can I find out what in my code is causing these warnings?  I'm not doing anything interesting with headers, not on purpose anyway.  http://pastebin.com/ydtwP1UH
12:32 osfabibisi denny: no idea - maybe it's worth linking your code though?
12:42 nic denny: What does that line of your Mojo::Headers say?
12:51 denny push @headers, "$name: $_" for @{$self->{headers}{lc $name}};
12:54 sue joined #mojo
13:03 neilhwatson joined #mojo
13:03 nic denny: If you run a test with MOJO_USERAGENT_DEBUG=1 you will see the headers; might reveal the answer
13:11 denny nic: thanks, I'll give that a go
13:13 denny Got it.  Thanks.
13:14 nic np
13:15 kaare joined #mojo
13:28 necrophcodr joined #mojo
13:28 necrophcodr anyone in here familiar with virtualmin pro?
13:29 necrophcodr i'm currently working on a script to determine what sites return HTTP/1.1 500
13:29 necrophcodr http://0bin.net/paste/yFBC526HyzpWbDCJ#y+reUckNT4IcJsDkGt-yxUrdZ0w/Ifood3ExB7hU2/X
13:29 necrophcodr this is how far i've gotten for now.
13:29 trone_ joined #mojo
13:30 necrophcodr however, it appears that *NO* sites return 500 at all, on any of our servers, which i know to be not true.
13:30 necrophcodr clearly i am doing something wrong, but i cannot figure out exactly what it is.
13:32 bpmedley necrophcodr : Do you have to stop the IOLoop?
13:32 necrophcodr bpmedley: i am not sure. i am not entirely sure how the Mojo::UserAgent perl module works yet, although from what i can tell, it is pretty good.
13:33 necrophcodr i mean the application terminates just fine when all requests have been processed.
13:39 bpmedley necrophcodr : I wonder if the app terminates when after a timeout.
13:40 necrophcodr bpmedley: i have no idea. is there any way to check that?
13:42 bpmedley Before doing that, you might try something like this: $ MOJO_USERAGENT_DEBUG=1 /opt/perl script.pl 2>&1 | grep '^HTTP/1.1'  and verify the status codes
13:44 necrophcodr there's no output from that command
13:45 bpmedley $ MOJO_USERAGENT_DEBUG=1 /opt/perl script.pl <-- What about something like this?
13:45 necrophcodr I get a lot of outputs like the following:
13:45 bpmedley Pastie might be better
13:45 necrophcodr it's just two lines, it'll be fine
13:46 necrophcodr alright, appearantly not, just a sec
13:46 necrophcodr -- Connect 225b98a0696e087a3bc59b31ac78ca3f (http://xstraw.dk.web4.redhost.dk:80)
13:46 necrophcodr -- Blocking request (http://energyplus.dk.web4.redhost.dk)
13:46 necrophcodr that's all the output
13:46 bpmedley Can you access the URLs via curl?
13:47 necrophcodr well, yes, but i do get a 403
13:47 necrophcodr which should be fine
13:48 bpmedley Why does your output say "Blocking request"?  It seems your script is non-blocking?
13:49 necrophcodr yeah, apologies, i changed it around to this:
13:49 necrophcodr http://0bin.net/paste/kFkGK+N1BPgwxCMI#DLuE9uOuki82o+Cd849hjitSEAjOcHQ0U3CK+VEkXwc
13:49 necrophcodr which is roughly the same
13:49 bpmedley necrophcodr : They are _not_ equivalent.
13:49 necrophcodr blocking or non-blocking doesn't matter to me, all that matters is using Mojo::UserAgent to retrieve the return code
13:50 bpmedley Hrmm
13:50 necrophcodr i am aware of the difference between blocking and non-blocking, and i know they're not the same
13:50 necrophcodr the reason i changed it to blocking was to assure myself that it wasn't an issue with the non-blocking method.
13:51 necrophcodr (especially considering the rest of the code is just good-ol' blocking i/o)
13:51 bpmedley Will you try a well known URL like google?
13:52 necrophcodr i'll have to rewrite the script a bit, but sure
13:53 necrophcodr works just fine
13:53 bpmedley Could be a proxy setting or somesuch?
13:53 necrophcodr there's no proxy on the servers or anything, no
13:54 bpmedley http://energyplus.dk.web4.redhost.dk <-- Will you hard code this?
13:54 necrophcodr it is pretty much just an apache2 setup on ubuntu, where each domain has a $DOM.$(hostname -f) serveralias
13:54 necrophcodr http://0bin.net/paste/uUmVe9EokEAsf2wf#GluYxamc5HrEj30nY8KDNls5qEiEENHf1JJltkfH83E
13:54 necrophcodr output from google
13:55 necrophcodr oh, useful information might be that i'm using perl 5.18.2, the latest on ubuntu 14.04.3
13:55 bpmedley Dunno.  Do you have cpan-outdated?
13:56 necrophcodr i do not, i'll look it up
13:56 bpmedley Perhaps, we can assume your perl env is fine.
13:56 necrophcodr Mojo::UserAgent is a recent installation too
13:57 bpmedley Will you hard code the energyplus.dk URL instead of google?
13:57 necrophcodr i will
13:57 necrophcodr i won't post the entirety of the output, but i'll post what is readable
13:58 necrophcodr result:
13:58 necrophcodr http://0bin.net/paste/LqHqIhCLMd1EN45i#UJ3f7ScQzzz4nwB5UIuWCr6fzvfYuCEi8VqsnHRe2ur
13:58 necrophcodr script:
13:58 necrophcodr http://0bin.net/paste/vYAH917BSwa0fkbf#Kle0Li2JFuugUE0PII8TeU9LhLn3Nex3teU3AF9MUtv
13:59 bpmedley necrophcodr : Try chomping $dom.
13:59 necrophcodr bpmedley: shit, that's a good idea, input sanitation and everything
13:59 zivester joined #mojo
14:00 pink_mist chomp(my @doms = `...`); instead of the s/\n// inside the loop perhaps?
14:00 bpmedley pink_mist : Is the issue $dom or @doms?  (or both)
14:00 pink_mist and chomp(my $dom = `...`) as well
14:00 necrophcodr pink_mist: yeah, i didn't think of chomp.
14:00 pink_mist could be both, could be either
14:01 necrophcodr i've done $dom and it worked, but i'll do it with $_ too (would it work just with @doms? really?)
14:01 pink_mist probably not an issue with @doms since he did have the s/\n// for @doms ... but I think that's a less than stellar idea
14:02 necrophcodr seems to work with @doms. perl is one smart fucker.
14:03 bpmedley necrophcodr : You might try looking into Mojo::IOLoop->delay for this.
14:03 necrophcodr i'll use my brain and eyes and read a bit harder next time. thanks a ton for your help!
14:12 henq hi, newbie question: if more than 1 route matches a request (i know, prob. bad practice) , which route gets to thandle the request? does the order in ihich routes are defined play a role (seems not).
14:13 bpmedley henq : http://mojolicious.org/perldoc/Mojolicious/Guides/Routing#Minimal-route
14:19 AlabamaPaul joined #mojo
14:32 Grinnz wow, that pastebin is trying to load flash for some reason
14:32 cpan_mojo JSON-Validator-0.65 by JHTHORSEN https://metacpan.org/release/JHTHORSEN/JSON-Validator-0.65
14:32 haarg Grinnz: for the copy to clipboard link
14:33 Grinnz ah
14:34 henq thx bpmedley. i got route /login defined before /:wiki , yet /login path is handled by /:wiki route , it seems; maybe a typo somewhere. thx again.
14:49 gryphon joined #mojo
14:51 bpmedley http://www.postgresql.org/about/news/1636/
14:53 Grinnz \o/
14:57 sri \o/
15:13 jberger gah, how refreshing to see an actual release announcement
15:14 sri ;p
15:15 cstamas hi
15:15 jberger cstamas: o/
15:15 cstamas i have the mojo headers object, how can i access a particular header
15:15 cstamas ?
15:16 jberger $headers->header('name')
15:16 cstamas ah, thx let me check :)
15:16 jberger pretty prominently documented of course ;-P
15:17 cstamas jberger: sorry, sometimes i am not that efficient following the docs :)
15:17 cstamas so from Mojolicious::Lite it would be like $c->req->headers->header('name') right?
15:18 sri maybe someone on the core team can get their employer to let them work on this for a few months? https://github.com/kraih/mojo/issues/423#issuecomment-169685551
15:18 jberger cstamas: yes
15:18 cstamas thx
15:18 jberger sri: I'm going to add an example of ->header to the synopsis in Mojo::Headers
15:19 jberger with all the specialized usages, the header method kinda gets lost
15:19 sri jberger: you're the pumpkin
15:20 bpmedley jberger : https://github.com/kraih/mojo/compare/master...brianmed:issue_882 <-- Thoughts on this?
15:21 jberger I'm probably not the best one to evaluate that
15:21 jberger I do think we stopped using localhost in favor of 127.0.0.1 at some point but perhaps that's not possible for this test
15:22 sri yea, the tests seem to be a no go portability wise
15:22 bpmedley sri : How can I improve them?
15:22 sri the cert seems to be missing from your branch as well
15:23 bpmedley I can add the certs real quick.. one sec..
15:23 sri and a description how to generate the cert
15:24 sri there is also no ca cert used by the user agent or tests for errors, which makes me suspicious
15:24 sri https://github.com/kraih/mojo/issues/882#issuecomment-169543994
15:25 bpmedley I definitely need better failing test(s)
15:26 sri there is already a 127.0.0.1 cert you can use
15:26 sri so, i would expect one additional cert for a domain like example.com
15:26 * jberger gives up trying to stash his local mojo changes
15:26 sri the CONNECT proxy would be used to establish the connection, so you don't have to mess with dns
15:26 * jberger makes another clone for day-to-day work while the refactor code is in play
15:27 Grinnz thats what branches are for :P
15:27 bpmedley I did visually inspect in chrome and got different hosts in the "Certificate information" tab.
15:28 bpmedley Ok, I'll google around for how to use a CA cert for testing.
15:29 sri bpmedley: https://github.com/kraih/mojo/blob/master/t/mojo/ioloop_tls.t#L13
15:29 sri those are the current instructions for how to regenerate the certs
15:29 bpmedley Ok, sweet.  I'll give it a go.
15:30 bpmedley Hopefully by tomorrow, no guarantees.
15:30 sri you just make a second server cert like that and you have all you need
15:31 bpmedley Understood; however, I'm piddling around with some other code right now.
15:31 bpmedley Also, $ork will start soon.
15:34 jberger sri / others: I'm having a hard time phrasing the lead-in comments, thoughts?
15:34 jberger http://pastie.org/10676187
15:34 jberger does that make sense/feel consistent with other docs?
15:36 bpmedley Methods for common headers <--> Accessors for common headers ??
15:36 sri jberger: # Parse and # Build are used all over the place
15:37 jberger bpmedley: they are accessor-like methods, not true accessors
15:37 bpmedley Understood
15:37 jberger which gave me pause documenting them that way
15:38 jberger sri: ok I could leave the parse and build alone and make a new section for read/retrieve/access?
15:39 jberger or I could just leave the comments alone and make the "code" changes I made
15:39 jberger maybe that's the easiest
15:42 bpmedley jberger : Currently, imo, by having ->content_length and ->content_type in the # Build section makes things confusing.
15:42 jberger bpmedley: the setters or the getters?
15:43 bpmedley say $headers->content_length  <-- This is confusing in the # Build section (to me).
15:45 sri pink_mist: fun fact, you actually do need threads to do async sockets right :)
15:45 sri as in you need a module like Net::DNS::Native, which uses a thread pool
15:47 sri since there's no native async version of getaddrinfo
15:48 pink_mist oh, that's less than awesome :/
15:48 Grinnz_ https://metacpan.org/pod/Net::LibAsyncNS is another option, but it doesn't bundle the required library atm
15:50 sri when you have Net::DNS::Native installed, your mojolicious apps will actually use 6 threads
15:50 cpan_mojo Mojolicious-Plugin-NYTProf-0.19 by LEEJO https://metacpan.org/release/LEEJO/Mojolicious-Plugin-NYTProf-0.19
15:53 sri bpmedley: may i critique your perl style a bit too?
15:53 bpmedley Sure
15:54 bpmedley It's a bit more verbose and free form than you like, I wager.. :)
15:54 sri the "$SSL_cert_file //= {};" doesn't serve a purpose, "$SSL_cert_file->{$1} = $query->param($name);" just works
15:54 sri same for the "next;", you can just use an elsif
15:55 bpmedley Awright, I sometimes can't remember when uninitialized warnings happen and when autovivification kicks in.
15:55 bpmedley Ok, cool.  I'll try to remember to make those changes.
15:58 sri i still think i can golf all that down to like 4 lines though ;p
16:17 odc joined #mojo
16:30 zivester joined #mojo
16:34 punter joined #mojo
16:34 cpan_mojo Galileo-0.040 by JBERGER https://metacpan.org/release/JBERGER/Galileo-0.040
16:37 jberger a causality of the list context behavior removal, the my @names = $c->param; one though so I missed it
16:37 jberger ^^
16:40 bpmedley joined #mojo
16:44 * sri kinda wants to hack on something again
16:46 miller joined #mojo
16:48 jberger sri: wanna see my most recent failed porting attempt?
16:48 jberger perhaps you can figure out what's missing?
16:54 sri bpmedley: hahaha, exactly 4 lines! https://gist.github.com/anonymous/8c3c6e4fd9fce80fcf59
16:55 bpmedley Awesome.  If you want, go crazy with the edits.  I won't be able to add much for several hours.
17:05 juikuen joined #mojo
17:21 sri jberger: sure
17:21 dod joined #mojo
17:30 good_news_everyon joined #mojo
17:30 good_news_everyon [mojo] jberger pushed 1 new commit to channel_jberger_4: http://git.io/vuiJV
17:30 good_news_everyon mojo/channel_jberger_4 a1cc9fc Joel Berger: move lots of client functionality over to the channel
17:30 good_news_everyon left #mojo
17:31 jberger sri: the problem is related to the data elements on the tx
17:31 jberger my first attempt tried to move them all to the channel
17:31 jberger then I realized I couldn't move {state} because the transaction needed to know about it (and publish it to consumers)
17:31 jberger and then there's the events
17:32 jberger anyway when you run the tests its the keep alive that complains, which tells me that something is missing in the finish/close logic
17:32 jberger and that keep-alive hates me
17:36 jberger I closed signatures 3/6 down can't pass a vote right
17:45 jberger sri: note that I queried you
17:47 sri yea, i guess signatures failed
17:48 sri tempire: you dropped out of #mojo-core again
17:53 PryMar56 joined #mojo
17:55 sri ok, i have a test scenario for server side SNI
17:56 sri it's a little crazy
18:06 good_news_everyon joined #mojo
18:06 good_news_everyon [mojo] kraih pushed 1 new commit to master: http://git.io/vuiCP
18:06 good_news_everyon mojo/master a2469e6 Sebastian Riedel: add SNI support to all built-in web servers
18:06 good_news_everyon left #mojo
18:09 sri i've used the CONNECT proxy so i don't have to mock anything
18:09 sri it just doesn't care what you request, and always conencts you to the local test server
18:10 sri so the user agent tells your local test server with SNI that it would like the cert for example.com
18:10 sri the user agent compares the hostnames and is happy
18:11 sri the user agent thinks it is connected to example.com the whole time
18:12 bpmedley Awesome sauce included.. :)
18:13 Kogurr joined #mojo
18:13 jberger the proxy is run over ipv6?
18:14 jberger oh but the whole test file already guards for that
18:14 jberger cool
18:14 PopeFelix joined #mojo
18:17 genio sri: Pg 9.5 released today \o/
18:17 genio that's probably already been noted
18:31 sri a few times ;p
18:31 sri wow, my postgres 9.5 tweet has 30 RTs :O
18:32 sri everybody loves postgres
18:34 sri hmm, guess it's only a matter of time until microsoft buys cyanogen and releases their own version of android
18:35 sri switching to cortana with cyanogen is a pretty bold move
18:47 good_news_everyon joined #mojo
18:47 good_news_everyon [mojo] kraih pushed 1 new commit to master: http://git.io/vuiof
18:47 good_news_everyon mojo/master e18a3b7 Sebastian Riedel: replace deprecated proxy method in Mojo::Message::Request with an attribute
18:47 good_news_everyon left #mojo
18:51 good_news_everyon joined #mojo
18:51 good_news_everyon [mojo] kraih pushed 1 new commit to master: http://git.io/vuiKH
18:51 good_news_everyon mojo/master 5832be6 Sebastian Riedel: a few more examples
18:51 good_news_everyon left #mojo
19:09 HtbaaPi joined #mojo
19:40 bjoernfan joined #mojo
19:40 sri jberger: re your branch, hard to spot anything without actually working on the branch
19:40 jberger yeah, its really tough
19:41 jberger my next plan is to make aliases between modules so that I can port smaller portions
19:41 jberger just as a porting mechanism
19:41 sri sounds reasonable
19:41 sri you're going to leave the state in the transaction?
19:42 jberger I'm not sure how not to
19:42 jberger so far the transaction doesn't know anything about the channel
19:43 sri what is left in the transactions that needs to know about the state?
19:43 jberger events
19:43 sri you mean you fire events when setting a state>
19:43 sri ?
19:43 jberger and a few things that check say 'finished'
19:43 sri that's not too bad
19:44 jberger there are such things already
19:44 sri i don't mind public transaction methods to fire events
19:44 jberger https://github.com/kraih/mojo/blob/master/lib/Mojo/Transaction.pm#L67-L71
19:44 jberger ok
19:44 sri yea, just replace those with a few simple public methods
19:45 sri like $tx->finish (just not really ->finish, since that's taken already for something else in the websocket tx ;p)
19:45 jberger there are a few places that check is_finished too
19:46 sri setting a finished flag on the tx should be fine
19:46 sri like when $tx->close is called (which fires the finished event or so)
19:46 jberger what about resume?
19:47 sri channel has to subscribe to the resume event?
19:47 jberger no, do we remove the resume method?
19:47 sri don't we need it?
19:47 jberger and if not, how does the transaction signal to the channel to resume
19:48 sri $tx->resume is called a lot from the controller
19:48 sri after rendering
19:48 jberger ok, so this needs dealing with
19:48 sri i think the channel has to subscribe to the resume event
19:48 jberger oh
19:48 jberger interesting
19:48 jberger yeah, that would work
19:48 sri daemon and user agent are subscribers already
20:16 sri might be worth taking a look at header safety again https://groups.google.com/forum/#!topic/mojolicious/FXuUZf-g7zU
20:42 sugar__ joined #mojo
20:49 hipp joined #mojo
20:54 ribasushi joined #mojo
21:03 mad_hatter joined #mojo
21:19 dod joined #mojo
21:23 PryMar56 joined #mojo
21:25 vanHoesel joined #mojo
21:42 miller_ joined #mojo
22:02 miller_ joined #mojo
22:07 good_news_everyon joined #mojo
22:07 good_news_everyon [mojo] kraih pushed 1 new commit to master: http://git.io/vuP7I
22:07 good_news_everyon mojo/master 329a070 Sebastian Riedel: more cert and key examples
22:07 good_news_everyon left #mojo
22:19 PopeF joined #mojo
22:34 zivester joined #mojo
23:25 sri looks like unix domain sockets are a bit controversial, maybe more folks would like to voice their opinion? https://github.com/kraih/mojo/issues/883
23:30 lluad I've used unix domain sockets for something vaguely similar. The only significant advantage over using tcp/localhost was the ability to use file permissions to control access.
23:30 lluad Performance was probably marginally higher, but for anything using http (or a perl framework) not enough that I'd care (or be able to measure it, I suspect).
23:31 lluad Seems like it'd be significant work for negligible benefit, unless someone has a compelling use case I'm missing.
23:32 lluad Bigger name space for listen ports too, perhaps. But 127.0.0.0/8 * 65000 isn't small.

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