| Time |
S |
Nick |
Message |
| 00:11 |
|
marcus |
hurrgh |
| 00:12 |
|
marcus |
sri: HALP |
| 00:12 |
|
marcus |
OAuth plugin works in async ways |
| 00:14 |
|
marcus |
and now my login action renders a 404 rather than waiting for the call |
| 00:14 |
|
marcus |
guess I need to pause it or something |
| 00:17 |
|
marcus |
uhm, seems there is no way to pause anything anymore. |
| 00:20 |
|
sri |
because everything is automatically paused |
| 00:21 |
|
marcus |
sri: I am trying to do $self->redirect_to('/new_event'); in a callback.. but it has already rendered a 404 because of no template file |
| 00:21 |
|
sri |
then you triggered automatic rendering |
| 00:22 |
|
|
GitHub198 joined #mojo |
| 00:22 |
|
GitHub198 |
mojo: master Sebastian Riedel * 725083e (1 files in 1 dirs): added tests - http://bit.ly/gSOghm |
| 00:22 |
|
|
GitHub198 left #mojo |
| 00:22 |
|
sri |
that test might answer it for you |
| 00:23 |
|
sri |
you likely named your action, not knowing it would trigger auto rendering |
| 00:23 |
|
sri |
response code turns it off again |
| 00:24 |
|
marcus |
aha |
| 00:24 |
|
marcus |
so I should probably do that in my plugin (setting the response code) |
| 00:24 |
|
marcus |
I named the action so that it would be easier to redirect to :) |
| 00:25 |
|
sri |
oh wait, possible bug found |
| 00:25 |
|
marcus |
hmm |
| 00:25 |
|
marcus |
if I set the code, it seems to have trouble with redirects |
| 00:26 |
|
sri |
this is highly experimental territory |
| 00:26 |
|
sri |
mixing blocking and async is really fucking hard |
| 00:26 |
|
marcus |
doh, of course it has problems with redirects if I put the code *after* the redirect. |
| 00:26 |
|
marcus |
sri: yepp |
| 00:26 |
|
marcus |
sri: still triggering autorendering |
| 00:26 |
|
sri |
thats why i said "oh wait" |
| 00:27 |
|
marcus |
:) |
| 00:28 |
|
|
MojoGuest39 joined #mojo |
| 00:28 |
|
MojoGuest39 |
From: http://news.ycombinator.com/item?id=1277067 (136 hits) |
| 00:28 |
|
|
MojoGuest39 left #mojo |
| 00:31 |
|
|
MojoGuest191 left #mojo |
| 00:32 |
|
marcus |
oh well. Good luck with the gruffalo hunt, sri. I am going to bed. Guess I'll be woken up in like 5 hours by a little terrrrist |
| 00:32 |
|
sri |
:) |
| 00:32 |
|
sri |
gonna add $self->delayed |
| 00:32 |
|
marcus |
sounds good |
| 00:33 |
|
marcus |
guess I can't release my plugin until the next mojo is out anyways then :) |
| 00:33 |
|
sri |
yea |
| 00:33 |
|
marcus |
sri: btw, do you have the url for the twitter access token url as well as the authorize one? |
| 00:33 |
|
sri |
nope |
| 00:34 |
|
marcus |
k |
| 00:39 |
|
sri |
ok, commit inc |
| 00:41 |
|
|
GitHub37 joined #mojo |
| 00:41 |
|
GitHub37 |
mojo: master Sebastian Riedel * 70a1233 (4 files in 4 dirs): added experimental delayed method to Mojolicious::Controller - http://bit.ly/higPRk |
| 00:41 |
|
|
GitHub37 left #mojo |
| 00:41 |
|
sri |
maybe a longer name like delayed_rendering would be a good idea |
| 00:45 |
|
marcus |
sri: yeah |
| 00:45 |
|
sri |
$self->render_later |
| 00:46 |
|
sri |
$self->delay_rendering |
| 00:46 |
|
sri |
render_later feels in line somehow |
| 00:46 |
|
sri |
$self->render_delayed |
| 00:46 |
|
marcus |
be_cool |
| 00:47 |
|
sri |
$self->have_a_drink |
| 00:47 |
|
sri |
$self->lay_back |
| 00:47 |
|
sri |
$self->relax |
| 00:48 |
|
sri |
relax would be kinda cool |
| 00:52 |
|
|
GitHub175 joined #mojo |
| 00:52 |
|
GitHub175 |
mojo: master Sebastian Riedel * be769ae (3 files in 3 dirs): renamed delayed method to render_later - http://bit.ly/heKM7X |
| 00:52 |
|
|
GitHub175 left #mojo |
| 00:53 |
|
|
curt left #mojo |
| 00:54 |
|
|
GitHub43 joined #mojo |
| 00:54 |
|
GitHub43 |
mojo: master Sebastian Riedel * 4c1970c (1 files in 1 dirs): pod cleanup - http://bit.ly/fnomvM |
| 00:54 |
|
|
GitHub43 left #mojo |
| 01:05 |
|
sri |
i actually thought about the render_later case above about a week ago, but got distracted before i could write a test :) |
| 01:11 |
|
perlrocks |
Twitter: "Having people actually learn #perl because of #mojolicious is really amazing and very motivating. :)" --kraih http://twitter.com/kraih/status/21372945611886592 |
| 01:12 |
|
perlrocks |
Twitter: "Wonder which beginner tutorials and books to link to from the #mojolicious documentation. #perl" --kraih http://twitter.com/kraih/status/21373299992825856 |
| 01:24 |
|
|
ispy_ left #mojo |
| 01:28 |
|
|
alnewkirk left #mojo |
| 01:51 |
|
tempire |
sri: How to access ioloop from within a plugin? |
| 01:55 |
|
|
GitHub55 joined #mojo |
| 01:55 |
|
GitHub55 |
mojo: master Sebastian Riedel * 98183b5 (1 files in 1 dirs): test cleanup - http://bit.ly/dGl4lk |
| 01:55 |
|
|
GitHub55 left #mojo |
| 01:55 |
|
sri |
tempire: that should answer the question |
| 01:56 |
|
sri |
the async client will always try to grab the right loop |
| 01:56 |
|
tempire |
huh |
| 01:56 |
|
tempire |
too easy |
| 01:56 |
|
sri |
you just check if it is running |
| 01:56 |
|
sri |
then you know if async is possible |
| 01:58 |
|
sri |
or $tx->connection might not even exist for most blocking environments |
| 02:02 |
|
tempire |
I'm using the following at the top of a test file with an embedded Mojolicious::Lite app: |
| 02:02 |
|
tempire |
http://pastie.org/1422419 |
| 02:02 |
|
tempire |
But I'm guessing it's not that simple, 'cause the test doesn't seem to be using IO::Socket::SSL |
| 02:03 |
|
tempire |
just INET |
| 02:03 |
|
perlrocks |
Twitter: "Test:: Mojo If you look properly decode I was I started wondering. Then move on and you look a little strange, Mojolicious charset is noticed that no output. Now." (ja) --kuusou_kyokusen http://twitter.com/kuusou_kyok[…]21386052056911872 |
| 02:04 |
|
sri |
tempire: didn't you look the the ioloop test case? |
| 02:04 |
|
tempire |
maybe not closely enough... |
| 02:04 |
|
sri |
https://github.com/kraih/mojo/[…]mojo/ioloop_tls.t |
| 02:04 |
|
sri |
the exact test case you need |
| 02:05 |
|
tempire |
Yeah, I have all that tested. Now I'm trying to test an ssl_auth plugin, so I want access to the Test::Mojo stuff. |
| 02:05 |
|
perlrocks |
Twitter: "The default charset = utf-8 Deta convinced that I'm supposed to like. Mojolicious:: Plugin:: Charset, but although you can specify or use, charset Did not it be better and I was always given me the security? What is XSS familiar." (ja) --kuusou_kyokusen http://twitter.com/kuusou_kyok[…]21386590358077440 |
| 02:05 |
|
* sri |
doesn't follow |
| 02:07 |
|
sri |
the translations are really bad, but i have a feeling that guy might be poisonous |
| 02:08 |
|
sri |
tempire: https://github.com/kraih/mojo/[…]_proxy_lite_app.t |
| 02:08 |
|
sri |
that test should have everything you need |
| 02:09 |
|
sri |
if you ignore the proxy stuff |
| 02:09 |
|
tempire |
ah |
| 02:09 |
|
tempire |
now I see what you were saying before |
| 02:09 |
|
* tempire |
follows |
| 02:09 |
|
sri |
:) |
| 02:10 |
|
|
mattastrophe left #mojo |
| 02:10 |
|
|
mattastrophe joined #mojo |
| 02:20 |
|
|
MojoGuest956 joined #mojo |
| 02:20 |
|
MojoGuest956 |
From: http://news.ycombinator.com/item?id=1277067 (137 hits) |
| 02:21 |
|
|
MojoGuest956 left #mojo |
| 02:40 |
|
|
curt joined #mojo |
| 02:48 |
|
|
josh left #mojo |
| 02:49 |
|
|
josh joined #mojo |
| 02:59 |
|
perlrocks |
Twitter: "Mojolicious:: Renderer should have default charset. Because Mojolicious:: Renderer already has default encoding # Mojolicious" (ja) --kuusou_kyokusen http://twitter.com/kuusou_kyok[…]21400041956577280 |
| 03:00 |
|
perlrocks |
Twitter: "Mojolicious:: Renderer:: render properly in "; charset = $ charset" I'm think of adding a right." (ja) --kuusou_kyokusen http://twitter.com/kuusou_kyok[…]21400450024611840 |
| 03:18 |
|
|
GitHub109 joined #mojo |
| 03:18 |
|
GitHub109 |
mojo: master Sebastian Riedel * fc641e3 (4 files in 4 dirs): improved charset handling in Mojolicious slightly - http://bit.ly/f4dWqC |
| 03:18 |
|
|
GitHub109 left #mojo |
| 03:19 |
|
|
mxey joined #mojo |
| 03:26 |
|
|
ispy_ joined #mojo |
| 03:26 |
|
mxey |
Hi. I started developing using Mojolicious and I really like it. I am wondering what you recommend to use for form validation? |
| 03:27 |
|
* sri |
waves |
| 03:27 |
|
* sri |
has no preference yet |
| 03:27 |
|
mxey |
Do you a list of options perhaps? :) |
| 03:28 |
|
mxey |
I need nothing fancy, I just have a user registration form. |
| 03:28 |
|
sri |
MojoX::Validator is rather popular from what i gathered |
| 03:29 |
|
mxey |
Looks good. |
| 03:29 |
|
mxey |
Thanks. |
| 03:54 |
|
|
curt left #mojo |
| 04:03 |
|
|
tholen42 joined #mojo |
| 04:05 |
|
|
tl left #mojo |
| 04:06 |
|
|
GeJ joined #mojo |
| 04:06 |
|
GeJ |
Good morning. |
| 04:24 |
|
tempire |
morning |
| 04:55 |
|
perlrocks |
Twitter: "@kraih I've enjoyed learning MORE perl and web development from the ground up because of #mojolicious and props for the help in #mojo" --ericrost http://twitter.com/ericrost/st[…]21429345696354306 |
| 07:36 |
|
|
kaare joined #mojo |
| 07:41 |
|
|
su-bzero[A] is now known as su-bzero |
| 07:42 |
|
|
verrens joined #mojo |
| 07:47 |
|
tempire |
I suppose I'm overlooking something simple. |
| 07:47 |
|
tempire |
http://pastie.org/1422876 |
| 07:47 |
|
tempire |
$loop is not the current ioloop. |
| 07:47 |
|
tempire |
it's just an empty blessed object |
| 07:52 |
|
perlrocks |
Twitter: "Mojo:: DOM I wonder what I used, was going to help test and useful. $ T-> tx-> res-> dom to take in, you can change the dynamic behavior of the test from the display of the current page. Convenient. # Mojolicious" (ja) --kuusou_kyokusen http://twitter.com/kuusou_kyok[…]21473823274569728 |
| 07:55 |
|
|
ispy_ left #mojo |
| 08:46 |
|
perlrocks |
Twitter: "MojoX:: Validator not you all. inflate, deflate feature is implemented you! From there, which I do not need to fork myself though I was lying implemented. # Mojolicious" (ja) --kuusou_kyokusen http://twitter.com/kuusou_kyok[…]21487602787950592 |
| 09:21 |
|
|
Alias_ left #mojo |
| 10:02 |
|
|
su-bzero_ joined #mojo |
| 10:04 |
|
su-bzero_ |
sri: What about community? Forum or something like that with voting for new features, plan for year,.. |
| 10:06 |
|
su-bzero_ |
People could help to improve mojolicious. |
| 10:07 |
|
su-bzero_ |
Now mojo doesn't have any community. maillist + irc are not enough. |
| 10:09 |
|
su-bzero_ |
And what will we do when you die? |
| 10:09 |
|
su-bzero_ |
Cause of beer and pizza. ;)) |
| 10:12 |
|
su-bzero_ |
Anyway community is very good for opensource software. |
| 10:13 |
|
su-bzero_ |
Only you will be able to commit, but allow people vote for features and etc. |
| 10:13 |
|
su-bzero_ |
Need more interactively framework. ;) |
| 10:15 |
|
* su-bzero_ |
.oO (Stop!) |
| 10:15 |
|
su-bzero_ |
;) Sorry. Just thoughts. |
| 10:18 |
|
lee7 |
methinks irc is Google Wave done right ;-P |
| 10:19 |
|
lee7 |
and perl culture always was irc-centric |
| 10:20 |
|
su-bzero_ |
It's so hard to read logs. |
| 10:20 |
|
su-bzero_ |
You should ask some questions again and again... |
| 10:23 |
|
lee7 |
then you stop asking, find an answer yourself and add it to Wiki :) |
| 10:27 |
|
su-bzero_ |
lee7: Yes, I know. No problems with questions. Problem is nobody could help to improve mojo, only sri could do it (as he thinks). |
| 10:29 |
|
lee7 |
benevolent dictatorship model has many advantages ;-| |
| 10:29 |
|
su-bzero_ |
Mojo is nice, sri did a good work. But this model could be improved too. ;) |
| 10:30 |
|
su-bzero_ |
What about voting for features? |
| 10:30 |
|
su-bzero_ |
It will be very usefull for project. |
| 10:30 |
|
lee7 |
what about voting for physical laws? it could be very useful too |
| 10:30 |
|
su-bzero_ |
Deputates vote. |
| 10:31 |
|
su-bzero_ |
People elect deputates. |
| 10:32 |
|
su-bzero_ |
No problems with current mojo model, but it will be better to do it better. ;)) |
| 10:32 |
|
lee7 |
with github you actually live in everett's multiverse, you CAN fork it :) |
| 10:33 |
|
su-bzero_ |
It's so hard to maintain this code. You should know. ;) |
| 10:33 |
|
su-bzero_ |
Only for sri's mind it's available. |
| 10:34 |
|
yko |
su-bzero_: using mojolicious you already elected sri as maintainer, so he voting for physical laws for you |
| 10:34 |
|
yko |
i'm sincerely as you to stop, really |
| 10:35 |
|
su-bzero_ |
Ok. |
| 10:35 |
|
lee7 |
acually, sri's code is shaping my mind, so far :) |
| 10:35 |
|
yko |
thank you |
| 10:35 |
|
su-bzero_ |
yko: Thank you for stopping me. |
| 10:35 |
|
su-bzero_ |
yko: You have two faces. ;)) |
| 10:35 |
|
su-bzero_ |
It were my thought, nothing else. |
| 10:35 |
|
su-bzero_ |
About mojo. |
| 10:36 |
|
|
su-bzero_ left #mojo |
| 10:42 |
|
rhaen |
moin |
| 10:42 |
|
rhaen |
happy new year! |
| 10:43 |
|
* rhaen |
just woke up. |
| 10:43 |
|
rhaen |
Oh, what a party! |
| 10:43 |
|
yko |
yeah! |
| 10:43 |
|
rhaen |
where are my pants? |
| 10:43 |
|
* rhaen |
looks around. |
| 10:44 |
|
rhaen |
time for a shower. |
| 10:44 |
|
yko |
lol. here's somebody pants, not sure it's yours, but whatever |
| 10:44 |
|
* yko |
takes pants hanging from chandelier. |
| 11:14 |
|
|
kaare left #mojo |
| 11:14 |
|
|
kaare joined #mojo |
| 11:16 |
|
|
daviddelikat joined #mojo |
| 11:30 |
|
|
verrens left #mojo |
| 11:56 |
|
perlrocks |
Twitter: "Blogs written - http://bit.ly/e5OSY5 - Mojolicious:: Lite in Data:: Model tried to use the" (ja) --nqounet http://twitter.com/nqounet/sta[…]21535177729646593 |
| 11:58 |
|
|
verrens joined #mojo |
| 12:04 |
|
|
Alias_ joined #mojo |
| 12:13 |
|
|
verrens left #mojo |
| 12:20 |
|
|
su-bzero_ joined #mojo |
| 12:29 |
|
|
bloonix left #mojo |
| 12:31 |
|
|
verrens joined #mojo |
| 12:39 |
|
|
verrens left #mojo |
| 12:45 |
|
|
s0va joined #mojo |
| 12:46 |
|
s0va |
buena nova! :) |
| 12:46 |
|
s0va |
question: i'd like to integrate mojo webapps into anyevent event framework... |
| 12:46 |
|
s0va |
i was reading source code |
| 12:47 |
|
s0va |
and i think that porting should be quite easy - some parts of ioloop should use anyevent infrastructure |
| 12:48 |
|
s0va |
i've seen AnyEvent::Mojo on cpan but it looks that is not working with current versions (>= 0.99936) of mojolicious |
| 12:50 |
|
s0va |
so my question is: if i port ioloop's functionality to use AnyEvent loop, would mojo work without any major changes? |
| 13:18 |
|
marcus |
sri: Added render_later and depends on 1.01 to OAuth plugin. |
| 13:20 |
|
marcus |
s0va: |
| 13:20 |
|
marcus |
x' |
| 13:58 |
|
|
curt joined #mojo |
| 14:06 |
|
|
daviddelikat left #mojo |
| 14:39 |
|
|
kvorg left #mojo |
| 14:52 |
|
|
ispy_ joined #mojo |
| 15:04 |
|
s0va |
marcus: ? |
| 15:06 |
|
yko |
he must be fall asleep right on the keyboard :) |
| 15:10 |
|
s0va |
:) |
| 15:11 |
|
s0va |
main motivation for porting mojo to anyevent is performance :) |
| 15:11 |
|
s0va |
simple mojo application using io::epoll does ~ 1100 simple req/s using keepalive... |
| 15:12 |
|
s0va |
simple anyevent::httpd based application does ~ 3100 reqs/s using keepalive |
| 15:13 |
|
s0va |
+ it seems that EV module implementation uses different approach by scheduling events. if you strace mojo process it does epol_something() + gettimeofday() every 0.25 seconds, EV's implementation does basically nothing just calls epoll_wait |
| 15:13 |
|
s0va |
nothing is called until something actually happens. |
| 15:14 |
|
|
daviddelikat joined #mojo |
| 15:14 |
|
s0va |
and by using anyevent you get full eventloop for free too :) |
| 15:19 |
|
yko |
i had a dream tonight that i found a way to decrease mojolicious memory consumption ~twice |
| 15:20 |
|
s0va |
yko: did you notice any leaks ... |
| 15:20 |
|
s0va |
becouse my mojo app grows over time... |
| 15:20 |
|
s0va |
i need to restart it once per week |
| 15:20 |
|
yko |
that should be your leaks, not mojolicious |
| 15:21 |
|
s0va |
yeah ... |
| 15:21 |
|
s0va |
if i strace process or inspect it using lsof... |
| 15:21 |
|
s0va |
there is nothing weird to see |
| 15:21 |
|
s0va |
i'm not allocating any buffers |
| 15:21 |
|
yko |
s0va : why don;t you try hypnotoad: it uses pool of daemons and shut down each after it serves n requests. (1000 by default) |
| 15:21 |
|
s0va |
i want singleprocess daemon... |
| 15:22 |
|
yko |
ah. i understand you |
| 15:22 |
|
s0va |
the only weird thing that could leak memory... |
| 15:22 |
|
s0va |
is using eval on read string... |
| 15:22 |
|
yko |
usually cross-links is what causes memory leaks |
| 15:23 |
|
s0va |
i get hashref in string and using eval i convert it into perl hashref... |
| 15:23 |
|
s0va |
is there any tool to inspect perl's memory heap? |
| 15:23 |
|
s0va |
something like jprofiler for java... |
| 15:24 |
|
yko |
Devel::Cycle is what i'm use to find leaks |
| 15:24 |
|
s0va |
i'll look into that. |
| 15:24 |
|
s0va |
i'd like to connect somehow to running process and inspect it's heap... |
| 15:25 |
|
yko |
i never go so deep. Devel::Cycle is quite enough for me |
| 15:44 |
|
marcus |
s0va: why do you want a singleprocess daemon? |
| 15:44 |
|
marcus |
s0va: hypnotoad is great for long running requests. |
| 15:46 |
|
|
bloonix joined #mojo |
| 15:48 |
|
perlrocks |
Twitter: "Started a pet project with Mojolicious, It's so easy to use and I was productive within a few minutes . Very cool #mojolicious" --mkirank http://twitter.com/mkirank/sta[…]21593623036428288 |
| 16:04 |
|
|
dotan joined #mojo |
| 16:11 |
|
s0va |
marcus: i *need* minimal memory footprint and i don't want to use multiple cache instances :) |
| 16:12 |
|
s0va |
plus i don't want to use some kind of ipc shared memory data sharing |
| 16:12 |
|
s0va |
my code is completely non-blocking therefore i don't see any advantage in using preforked server :) |
| 16:24 |
|
marcus |
s0va: what about 0 downtime code upgrades? |
| 16:25 |
|
marcus |
s0va: and are you running your stuff on a single-core cpu? |
| 16:37 |
|
sri |
s0va: you might be misunderstanding a few things |
| 16:38 |
|
sri |
s0va: how did you test AnyEvent::HTTPD with mojolicious? |
| 16:39 |
|
sri |
assuming it will just magically increase req/s by over 200% is bullshit |
| 16:40 |
|
sri |
for micro benchmarks use examples/microhttpd.pl |
| 16:40 |
|
sri |
1100 req/s is a lready an unrealistic number for any serious application |
| 16:43 |
|
sri |
s0va: using multiple processes has many advantages... or why dod you think nginx does it too? |
| 16:43 |
|
sri |
*-d |
| 16:44 |
|
sri |
s0va: bottom line, if all you're after is "performance", don't even start with an interpreted language |
| 16:45 |
|
sri |
you either optimize for programmers or cpu, not both |
| 16:45 |
|
marcus |
unless your programmers are cpus... |
| 16:45 |
|
marcus |
:) |
| 16:45 |
|
sri |
*mind blown* |
| 16:48 |
|
sri |
to put things in perspective, reddit handles 80 req/s per server during peek hours, and thats pretty damn good for a real world app |
| 16:50 |
|
sri |
*peak |
| 16:54 |
|
sri |
for the record, examples/microhttpd.pl does 25k req/s |
| 16:55 |
|
sri |
surely it has to be the best production http server... or maybe not ;p |
| 16:56 |
|
marcus |
sri: the best production server for my awesome upcoming site helloworld.com |
| 16:56 |
|
sri |
! |
| 16:56 |
|
perlrocks |
Twitter: "Mojolicious - Web framework for Perl http://www.findbestopensource.[…]oduct/mojolicious # opensource" (id) --bestopensource http://twitter.com/bestopensou[…]21610904839782400 |
| 16:57 |
|
marcus |
findbestopensource seems pretty spammy |
| 16:58 |
|
sri |
su-bzero: this is a meritocracy, if you want to decide you have to contribute |
| 16:59 |
|
sri |
su-bzero: take rhaen for example, he is the most qualified to maintain the wiki, so he has taken it over |
| 16:59 |
|
perlrocks |
Twitter: "Mojolicious - Web framework for Perl http://www.findbestopensource.[…]oduct/mojolicious # opensource http://bit.ly/dQLmwy" (id) --FOSSwiki http://twitter.com/FOSSwiki/st[…]21611618555133953 |
| 17:17 |
|
tempire |
sri: Alright, so I've got my tests for specifying an ssl cert & running it through my Mojolicious::Plugin::SslAuth, which I gleaned from the https://github.com/kraih/mojo/[…]_proxy_lite_app.t you showed me. |
| 17:18 |
|
tempire |
My problem is that I can't seem to access the current ioloop from within the plugin. I've tried $self->client->async->ioloop & $self->client->ioloop, like the other test files you mentioned. |
| 17:18 |
|
tempire |
halp! |
| 17:19 |
|
* tempire |
waits patiently |
| 17:23 |
|
tempire |
Also, I'm trying to think of a simple way to specify certificates with Mojo::Client. I suppose print $client->get("http://kraih.com:$port:$cert:$key"') could work, to keep it consistent with --listen |
| 17:26 |
|
tempire |
Seems clunky to me, since ":$cert:$key" aren't actually part of location, which invalidates the concept of a urL |
| 17:27 |
|
tempire |
hmm, I wonder if $client->get("http://cert:key kraih.com") would make sense. The client could detect that cert:key were files and operate accordingly. |
| 17:27 |
|
tempire |
That's kind of clunky as well. |
| 17:28 |
|
moritz |
would that still do a normal HTTP request? |
| 17:28 |
|
moritz |
if not, I don't see why it should be doen with ->get("http://...") |
| 17:28 |
|
moritz |
if it's not the standard action, don't use the standard method |
| 17:29 |
|
tempire |
you're right. |
| 17:30 |
|
tempire |
->get('http://kraih.com' => {Connection => 'close'} => 'Hi!' => "cert:key" => sub {...}) I guess |
| 17:30 |
|
marcus |
what about adding a separate method for it? |
| 17:30 |
|
marcus |
or setting the certificate on the client before the get... |
| 17:31 |
|
tempire |
->cert($cert)->key($key)->get(...) ? |
| 17:31 |
|
tempire |
or ->ssl("$cert:$key")->get(...) |
| 17:32 |
|
marcus |
I like the first one better. |
| 17:32 |
|
marcus |
but sri probably has some opinions once he's done recharging or whatever |
| 17:33 |
|
tempire |
ah, I meant ->get("https://cert:keykraigh.com"), earlier |
| 17:33 |
|
tempire |
but the ->cert->key->get is a much better idea |
| 17:33 |
|
yko |
async get() should return client itself |
| 17:34 |
|
yko |
what ->cert() will return? |
| 17:34 |
|
tempire |
$self (the client) |
| 17:35 |
|
tempire |
Same with ->key |
| 17:35 |
|
yko |
so, after get() any other requests with that client will use same certificate? |
| 17:35 |
|
tempire |
You would have to clear those values |
| 17:35 |
|
yko |
oh... |
| 17:36 |
|
tempire |
$self->cert($cert)->key($key)->get(...)->cert("")->key("")->get(...) |
| 17:36 |
|
tempire |
But in that case, it seems like you'd be better off making another client object |
| 17:36 |
|
yko |
i think there should be a sub that will prepare tx |
| 17:36 |
|
* sri |
likes |
| 17:36 |
|
yko |
but i'm not familar with specs though :( |
| 17:36 |
|
* sri |
suggests cert_file and key_file though |
| 17:37 |
|
sri |
since some day support for in memory certs might be added |
| 17:37 |
|
tempire |
I was going to make ioloop recognize that shortly |
| 17:38 |
|
tempire |
tls_key & tls_cert can easily detect pem format or file |
| 17:38 |
|
tempire |
->cert & ->key would just be passed to ioloop, and could thereby take either a file or pem string |
| 17:38 |
|
tempire |
you could interchange as well |
| 17:39 |
|
sri |
sounds good, don't have to worry about that yet anyway |
| 17:40 |
|
sri |
IO::Socket::SSL only supported files last i checked |
| 17:40 |
|
tempire |
it supports strings as well |
| 17:40 |
|
sri |
:o |
| 17:40 |
|
tempire |
I haven't tested it, but the docs imply as such |
| 17:41 |
|
tempire |
SSL_key vs SSL_key_file |
| 17:41 |
|
sri |
hmm |
| 17:42 |
|
sri |
can you clean up the code for the built in certs while you are at it anyway? :) |
| 17:42 |
|
sri |
they are still written to files |
| 17:42 |
|
tempire |
that's the plan |
| 17:42 |
|
* tempire |
bows in service |
| 17:42 |
|
sri |
\o/ |
| 17:43 |
|
tempire |
but back to my question about ioloop earlier |
| 17:43 |
|
tempire |
halp! |
| 17:43 |
|
sri |
? |
| 17:43 |
|
tempire |
tempire: sri: Alright, so I've got my tests for specifying an ssl cert & running it through my Mojolicious::Plugin::SslAuth, which I gleaned from the https://github.com/kraih/mojo/[…]_proxy_lite_app.t you showed me. |
| 17:43 |
|
tempire |
tempire: My problem is that I can't seem to access the current ioloop from within the plugin. I've tried $self->client->async->ioloop & $self->client->ioloop, like the other test files you mentioned. |
| 17:44 |
|
sri |
that would be quite weird |
| 17:44 |
|
sri |
considering there are tests doing exactly that |
| 17:46 |
|
sri |
whole longpolling file is full of them |
| 17:47 |
|
tempire |
I'm using ioloop->connect to connect to the server, as opposed to Mojo::Client, maybe that's why ioloop isn't available |
| 17:49 |
|
tempire |
http://pastie.org/1423724 |
| 17:49 |
|
tempire |
does anything look inherently wrong in that? |
| 17:49 |
|
|
curt left #mojo |
| 17:49 |
|
sri |
oh, that changes everything |
| 17:50 |
|
|
curt joined #mojo |
| 17:50 |
|
sri |
new client instance is not using the ioloop singleton |
| 17:50 |
|
sri |
you would need an outer ioloop in that case |
| 17:50 |
|
tempire |
ah |
| 17:50 |
|
sri |
you have to manually make sure they are all in the same ioloop |
| 17:52 |
|
tempire |
so Mojo::IOLoop->new->connect(...) |
| 17:53 |
|
sri |
Mojo::IOLoop->singleton |
| 17:53 |
|
sri |
then $self->client->async->ioloop should work |
| 17:54 |
|
sri |
your problem is that you were setting up too much manually |
| 17:57 |
|
tempire |
The point is to use the same ioloop for both the server & client request, right? |
| 17:58 |
|
tempire |
hence the use of the singleton |
| 18:03 |
|
tempire |
so then this should work: http://pastie.org/1423758 |
| 18:04 |
|
yko |
maybe, at line 3 should be $ioloop? |
| 18:06 |
|
sri |
thats exactly the point |
| 18:06 |
|
sri |
but $self->client usually starts with a random ioloop |
| 18:07 |
|
sri |
so you rely on $self->client->async to detect the outer loop |
| 18:07 |
|
yko |
also client may need to have app defined, no? |
| 18:07 |
|
|
dku joined #mojo |
| 18:07 |
|
sri |
which in turn usually relies on Mojo::IOLoop->singleton |
| 18:08 |
|
sri |
yko: unrelated here i think |
| 18:08 |
|
tempire |
yko: in this case, there is no client, I'm only using it to generate an ioloop |
| 18:08 |
|
tempire |
(no Mojo::Client, I mean) |
| 18:08 |
|
|
dotan left #mojo |
| 18:08 |
|
sri |
tempire: that paste makes no sense though |
| 18:09 |
|
sri |
my $loop = Mojo::Client->new->ioloop; $loop = Mojo::IOLoop->singleton; |
| 18:09 |
|
sri |
makes no sense at all |
| 18:09 |
|
sri |
the point is to have them all use the same |
| 18:10 |
|
tempire |
agreed. tried that, I'm just trying different variations based on your comments. |
| 18:11 |
|
tempire |
although it seems to me the use of $loop = Mojo::IOLoop->singleton be the same thing as not having that statement at all. |
| 18:11 |
|
* sri |
doesn't follow |
| 18:12 |
|
* tempire |
is awaiting the outcome of this, wherein he will be an IOLoop master |
| 18:12 |
|
sri |
well, this is not everyday ioloop stuff |
| 18:15 |
|
tempire |
so http://pastie.org/1423787 , combined with $self->client->async->ioloop (http://pastie.org/1423790) should give me the ioloop I'm looking for. |
| 18:16 |
|
sri |
no |
| 18:16 |
|
sri |
still totally wrong |
| 18:16 |
|
tempire |
awesome! |
| 18:16 |
|
sri |
;p |
| 18:16 |
|
sri |
my $loop = Mojo::IOLoop->singleton; |
| 18:16 |
|
sri |
thats how you start |
| 18:17 |
|
tempire |
well |
| 18:17 |
|
tempire |
freaking eh |
| 18:17 |
|
tempire |
it's like magic. |
| 18:17 |
|
sri |
singleton is the one magic loop |
| 18:18 |
|
tempire |
Ah, I was thinking incorrectly. I was assuming that Mojo::Client->new->ioloop would be recognized by singleton all throughout, because it was the only ioloop within the application. |
| 18:18 |
|
tempire |
But that was too much assumption. |
| 18:18 |
|
sri |
ye |
| 18:19 |
|
sri |
client always starts blocking with a fresh anonymous ioloop |
| 18:19 |
|
tempire |
it's so obvious looking at IOLoop.pm now |
| 18:19 |
|
tempire |
:s |
| 18:20 |
|
* tempire |
declares himself a master of all things |
| 18:20 |
|
sri |
the downside of defaulting to blocking, makes the hard test cases a little harder |
| 18:21 |
|
sri |
but this is about as hard as it gets |
| 18:22 |
|
yko |
btw, sri thank you again for pointing me on client.t for nested requests. |
| 18:22 |
|
sri |
sure thing |
| 18:23 |
|
yko |
i was fighting with windmills, because problem was not on my side, not on mojo side but on aspx side :s |
| 18:25 |
|
tempire |
Ha! |
| 18:25 |
|
tempire |
"Construct a new Mojo::IOLoop object. Multiple of these will block each other, so use singleton instead if possible." |
| 18:25 |
|
tempire |
everything falls into place |
| 18:25 |
|
sri |
\o/ |
| 18:25 |
|
yko |
or chain each against other :) |
| 18:26 |
|
tempire |
I want to add a $loop->connection($id) to get the contents of $loop->{_cs} |
| 18:26 |
|
tempire |
or $loop->handle($id) to get the IO::Socket:* would do as well |
| 18:26 |
|
sri |
$loop->handle sounds ok |
| 18:27 |
|
tempire |
I suppose that is more consistent with the other calls |
| 18:29 |
|
sri |
or wait |
| 18:29 |
|
sri |
tempire: do you really need the socket? |
| 18:30 |
|
sri |
how about $loop->remote_info($id) adding the cert too? |
| 18:31 |
|
tempire |
There are several IO::Socket::SSL calls I can foresee people needing. |
| 18:31 |
|
sri |
:S |
| 18:31 |
|
tempire |
I know, I know |
| 18:31 |
|
tempire |
actually, I may be misunderstanding IO::Socket::SSL docs |
| 18:32 |
|
tempire |
hold on... |
| 18:33 |
|
tempire |
Ok, the problem is there's more than peer_certificate |
| 18:33 |
|
tempire |
There's get_cipher & verify_hostname |
| 18:34 |
|
sri |
guess ->handle it is then |
| 18:34 |
|
tempire |
I guess those can be added to remote_info, though. |
| 18:35 |
|
tempire |
What's your main concern about exposing the handle? |
| 18:36 |
|
sri |
many unknowns |
| 18:37 |
|
sri |
people will depend on IO::Socket::SSL |
| 18:37 |
|
tempire |
Which could change at any time |
| 18:37 |
|
tempire |
And then you get blamed. |
| 18:37 |
|
* sri |
is still hoping for a more sane tls module to come along |
| 18:38 |
|
sri |
but that migth just be wishful thinking |
| 18:38 |
|
sri |
we might also have to use some ipv6 socket module |
| 18:38 |
|
sri |
which would change normal handles |
| 18:39 |
|
tempire |
The alternative is to include calls to ::SSL in remote_info, which still couples it tightly to mojolicious. |
| 18:39 |
|
sri |
maybe i'm overthinking it |
| 18:39 |
|
tempire |
But in that case, you'd have to change the whole TLS implementation anyway. |
| 18:39 |
|
sri |
just go with handle |
| 18:41 |
|
sri |
if you want to be nice you can add an ->isa check to your plugin |
| 18:41 |
|
tempire |
yeah |
| 18:41 |
|
tempire |
fail if not SSL |
| 18:41 |
|
tempire |
::SSL |
| 18:41 |
|
sri |
ye |
| 18:41 |
|
tempire |
It doesn't seem like {_cs} gets populated in ioloop.t. I'm guessing you have to have a listen & connection to get it. |
| 18:42 |
|
tempire |
ah, connect |
| 18:42 |
|
sri |
connections might require a tick |
| 18:43 |
|
tempire |
->one_tick? |
| 18:43 |
|
tempire |
That doesn't touch {_cs} |
| 18:43 |
|
sri |
you shouldn't have to do anything manually though |
| 18:44 |
|
sri |
dns lookup delays the actual connect i think |
| 18:44 |
|
sri |
by one tick with a 0 timer |
| 18:44 |
|
sri |
not sure why you worry about that though |
| 18:44 |
|
tempire |
I'm looking for the best way to test ->handle, which means I need something in {_cs} |
| 18:45 |
|
sri |
use a real test |
| 18:45 |
|
tempire |
I can test it in the tls tests, but it seems to make more sense to do it in ioloop.t |
| 18:45 |
|
sri |
ioloop.t doesn't do anything besides timers ;p |
| 18:45 |
|
sri |
ioloop_tls.t |
| 18:46 |
|
tempire |
k |
| 18:46 |
|
sri |
you could add a new test to ioloop.t though |
| 18:46 |
|
sri |
removing the tls parts |
| 18:48 |
|
tempire |
so...to be clear, you don't want the handle test in ioloop.t |
| 18:48 |
|
tempire |
but rather in ioloop_tls.t |
| 18:49 |
|
tempire |
seems inconsistent for tracking down tests later |
| 18:50 |
|
tempire |
oh, nevermind |
| 18:50 |
|
tempire |
I didn't read those last two statements |
| 18:51 |
|
sri |
:) |
| 18:52 |
|
sri |
i never added a basic socket test to ioloop.t because basically every lite_app test does the same thing already |
| 18:52 |
|
sri |
handle is a good reason to do it now though |
| 18:59 |
|
tempire |
Is there any time there wouldn't be a handle in an {_cs}->{$id}? |
| 19:00 |
|
sri |
if a handle exists it is there |
| 19:01 |
|
sri |
{_cs}->{$id}->{handle} of course, but i guess you mean that |
| 19:01 |
|
tempire |
yes |
| 19:02 |
|
tempire |
I'm just wondering if there's a situation where {_cs}->{$id} is a hashref but {_cs}->{$id}->{handle} does not exist. |
| 19:02 |
|
tempire |
but I'm seeing, no |
| 19:10 |
|
perlrocks |
Twitter: "Blogs written - http://bit.ly/gGuqfA - Mojolicious:: Lite in DBIx:: Skinny tried to use the" (ja) --nqounet http://twitter.com/nqounet/sta[…]21644574837448704 |
| 19:11 |
|
tempire |
wait a second. |
| 19:11 |
|
tempire |
It's just occurring to me that if I provide IO::Socket::SSL in my ssl_auth plugin, I'll have to support that as well. |
| 19:12 |
|
* tempire |
is starting to feel some of sri's pressure |
| 19:12 |
|
sri |
:) |
| 19:17 |
|
|
Akron joined #mojo |
| 19:36 |
|
tempire |
BOOSH! |
| 19:36 |
|
tempire |
ssl plugin working |
| 19:36 |
|
tempire |
and tested |
| 19:36 |
|
sri |
\o/ |
| 19:36 |
|
tempire |
now to attack ->cert & ->key |
| 19:37 |
|
tempire |
using $loop->connect manually is painful in comparison |
| 19:38 |
|
sri |
just add a tls_lite_app.t for testing the higher level api |
| 19:40 |
|
seb_ |
hi there, does someboy know a simple working example of websocket for mojolicious? All examples from the site did not function for diverses resons. |
| 19:41 |
|
sri |
how about the echo example on http://mojolicio.us? |
| 19:41 |
|
sri |
doesn't get much simpler |
| 19:41 |
|
s0va |
sri: about anyevent && stuff: i'd like to run mojo apps as part of anyevent webapp... |
| 19:42 |
|
sri |
s0va: thats not what you said before |
| 19:42 |
|
sri |
s0va: but you could write a replacement ioloop using anyevent if you wanted |
| 19:42 |
|
s0va |
sri: my question was: if i port ioloops functionality to use anyevent loop, is there a chance to run mojo apps without major rewrite of mojo? |
| 19:43 |
|
s0va |
as far i see, lots of mojo great stuff depends on ioloop. |
| 19:43 |
|
s0va |
i'm already very pleased with mojo's performance... |
| 19:44 |
|
sri |
$Mojo::IOLoop::LOOP = Mojo::IOLoop::Anyevent->new |
| 19:44 |
|
sri |
you might be able to do something like that |
| 19:44 |
|
s0va |
sri: server::daemon and mojo::client should work out of the box if anyevent compatible ioloop would be available in Mojo::IOLoop? |
| 19:45 |
|
sri |
sure, everything is only using public apis |
| 19:45 |
|
s0va |
i was thinking about something like inet6 glue module, which copies all symbols of IO::Socket::INET6 in place of IO::Socket::INET |
| 19:45 |
|
s0va |
great :) |
| 19:46 |
|
s0va |
anyevent also natively support TLS, DNS async resolution, timers, watchers and even working IPv6! |
| 19:46 |
|
sri |
not much to be gained there |
| 19:47 |
|
sri |
we have all that |
| 19:47 |
|
s0va |
sri: true, but the big gain would be anyevent integration :) |
| 19:47 |
|
sri |
there is also another alternative |
| 19:47 |
|
s0va |
running multiple mojo apps in single perl process etc |
| 19:47 |
|
sri |
we can already do that |
| 19:47 |
|
s0va |
true :) |
| 19:48 |
|
sri |
you could run anyevent next to the normal ioloop with an on_tick event |
| 19:48 |
|
s0va |
anyevent doesn't have one_tick() equivalent |
| 19:48 |
|
sri |
the other way around |
| 19:48 |
|
s0va |
... this is where we have a problem :) |
| 19:49 |
|
s0va |
sri: i just want to code nice gui for my existing daemon app using mojo |
| 19:49 |
|
sri |
Mojo::IOLoop->singleton->on_tick(sub {EV::loop EV::LOOP_NONBLOCK}); |
| 19:49 |
|
sri |
something like that |
| 19:49 |
|
s0va |
omg :) |
| 19:50 |
|
sri |
there is also a poe example in the wiki |
| 19:50 |
|
s0va |
yeah, i saw :) |
| 19:50 |
|
s0va |
fuck poe, and as i said some time ago: POE wheels for some weird reason block the entire process |
| 19:51 |
|
seb_ |
sri: what does the simple echo websocket example do? I have a websocket rocks on my localhost:80 (!), and links /title /echo /clock show me the text value {"list":[0]} |
| 19:51 |
|
sri |
the community around poe is nice |
| 19:52 |
|
seb_ |
(using google chrome) |
| 19:52 |
|
s0va |
sri: i'll try to implement lowlevel stuff you do in ioloop using anyevent; i think that both communities would benefit from it. |
| 19:52 |
|
sri |
seb_: you might want to read up on websockets first |
| 19:52 |
|
s0va |
anyevent's httpd really sucks donkey's balls compared to mojo::server::daemon |
| 19:52 |
|
sri |
they have a special javascript api you need to use |
| 19:53 |
|
sri |
s0va: sure, let me know if there are any small changes needed to make it easier |
| 19:53 |
|
s0va |
sri: in a week or so :) |
| 19:53 |
|
sri |
like $ENV{IOLOPP_CLASS} = 'Mojo::IOLoop::Anyevent' or so |
| 19:53 |
|
s0va |
sri: i also found that anyevent is really not very memory hungry... |
| 19:54 |
|
s0va |
sri: yeah, i was thinking about something like that to... |
| 19:54 |
|
s0va |
plus i found that full ssl/tls support is already written with certificate verification |
| 19:54 |
|
sri |
s0va: you might want to wait for tempire though, he's making some tls adjustments right now |
| 19:54 |
|
s0va |
without io::socket::ssl stuff |
| 19:55 |
|
sri |
heh, we'll have that soon |
| 19:55 |
|
s0va |
np, i'm regulary pulling git repo |
| 19:56 |
|
s0va |
anyevent also does stuff your way: it's all about closures :) |
| 19:56 |
|
* sri |
is more of a parallel event loop fan though |
| 19:56 |
|
sri |
i'm sure you could do cool things with on_tick, Coro and EV |
| 19:56 |
|
sri |
to increase responsiveness |
| 19:57 |
|
s0va |
i've done lots of poe programming; it's quite nice when you get used to it, but you hit the scaling barrier really fast |
| 19:57 |
|
s0va |
aham... EV's loop doesn't have tick, it doesn't wake up on regular basis... |
| 19:57 |
|
s0va |
and that's good :) |
| 19:58 |
|
s0va |
EV uses libev, which should be even more scalable than libevent |
| 19:59 |
|
s0va |
sri: however, i'll dig into it. i just wanted to be sure if ioloop implementation is core of mojo business :) |
| 20:00 |
|
sri |
"EV uses libev, which should be even more scalable than libevent" |
| 20:00 |
|
sri |
thats a rather weird phrase |
| 20:01 |
|
sri |
what makes it more "scalable" exactly? |
| 20:01 |
|
s0va |
http://libev.schmorp.de/bench.html |
| 20:01 |
|
sri |
ok, what has raw performance to do with scalability? |
| 20:01 |
|
s0va |
handling of more events/fds concurrently |
| 20:01 |
|
sri |
thats performance |
| 20:02 |
|
sri |
it does nto help with scaling |
| 20:03 |
|
sri |
for example the actor model is a solution for scalability |
| 20:04 |
|
seb_ |
I prefer living examples to learn. I will search some more. I do not understand from where come this "websocket test" page (wich is not in the code), who connect who, where do echo.websocket.org intervince. |
| 20:05 |
|
sri |
seb_: you need to undertsand the javascript side first |
| 20:05 |
|
sri |
then it will make sense ;) |
| 20:06 |
|
sri |
websockets are a lower level async message passing layer |
| 20:06 |
|
sri |
relatively unrelated to http |
| 20:07 |
|
sri |
s0va: if it's raw performance you're after then http://search.cpan.org/dist/Nginx-Engine/ might be the new king in town |
| 20:08 |
|
tempire |
Has anyone used that? |
| 20:08 |
|
sri |
it should be pretty much impossible to beat since all io happens in C space |
| 20:08 |
|
sri |
no io handles at all |
| 20:08 |
|
|
curt left #mojo |
| 20:08 |
|
sri |
not used it yet, but like the api |
| 20:09 |
|
|
curt joined #mojo |
| 20:09 |
|
sri |
the node.js comparison is quite correct, since it has no io handle concept either |
| 20:13 |
|
seb_ |
I stay confuse. Do you have some good ressources. Google geve all in it's reverse for websocket. Where should this be on the OSI model? |
| 20:13 |
|
sri |
seb_: https://github.com/kraih/mojo/[…]elloWorld.pm#L293 |
| 20:14 |
|
sri |
thats how the javascript code should look like |
| 20:14 |
|
sri |
for a minimal example |
| 20:15 |
|
sri |
the websocket protocol is on the same layer as http |
| 20:15 |
|
sri |
it's an alternative for real time communication |
| 20:15 |
|
seb_ |
thanx I will try this |
| 20:16 |
|
sri |
no need to rush learning websockets though |
| 20:16 |
|
sri |
they will break soon |
| 20:16 |
|
seb_ |
? |
| 20:16 |
|
sri |
the protocol is not done yet |
| 20:17 |
|
sri |
the next version of the spec is guaranteed to break every implementation |
| 20:20 |
|
|
bloonix left #mojo |
| 20:20 |
|
seb_ |
I will not wait to understand this :). I have tons of ideas to test. And it look like I have some road to walk before |
| 20:24 |
|
sri |
someone should combine nginx with v8 to kick node.js ass :) |
| 20:28 |
|
sri |
oh cool, new alien series this year! |
| 20:28 |
|
sri |
Falling Skies |
| 20:28 |
|
sri |
looks like a mix of walking dead and V |
| 20:29 |
|
yko |
oh ye? |
| 20:29 |
|
sri |
http://www.imdb.com/video/imdb/vi1185651481/ |
| 20:29 |
|
yko |
walking dead freaked out with idle for one year |
| 20:30 |
|
yko |
so powerfull start, so cool camera work and such a.... i dont know proper words in english :) |
| 20:32 |
|
sri |
wait, are you saying there will be no new walking dead episodes for a year? |
| 20:32 |
|
sri |
NOOOOOOOOOOOOOOOOOOOOO!!!! |
| 20:33 |
|
yko |
lol, you didn't know? |
| 20:33 |
|
* sri |
cries |
| 20:33 |
|
yko |
they said that they will rest for about a year |
| 20:43 |
|
|
seb_ left #mojo |
| 20:46 |
|
|
kvorg joined #mojo |
| 20:58 |
|
|
curt left #mojo |
| 20:59 |
|
|
curt joined #mojo |
| 21:00 |
|
|
baton8 left #mojo |
| 21:06 |
|
tempire |
ahhh! |
| 21:06 |
|
tempire |
http://planet.perl-magazin.de/ |
| 21:06 |
|
tempire |
We can't let the germans take over |
| 21:06 |
|
tempire |
It usually doesn't go well |
| 21:06 |
|
tempire |
vti is the only one who can save us |
| 21:07 |
|
* tempire |
listens for a trolololo |
| 21:12 |
|
|
baton8 joined #mojo |
| 21:14 |
|
|
achromic joined #mojo |
| 21:15 |
|
sri |
trolololo |
| 21:23 |
|
|
baton8 left #mojo |
| 21:23 |
|
curt |
all, I'm working through the routing example on git... https://github.com/kraih/mojo/[…]for-non-lite-apps. I'm having trouble with the new_form portion of the example... |
| 21:24 |
|
curt |
<%= form_for '/cities' => (method => 'post') => begin %> is routing to http://localhost:3000/cities/cities |
| 21:25 |
|
curt |
but changing the example to <%= form_for 'http:///cities' => (method => 'post') => begin %> resolves the problem. |
| 21:28 |
|
|
baton8 joined #mojo |
| 21:32 |
|
yko |
hm |
| 21:34 |
|
yko |
problem lies here: http://bit.ly/i0tZpj |
| 21:34 |
|
yko |
but i'm unsure is it a problem or no. not sure to what this commit http://bit.ly/gpdx5v was related |
| 21:35 |
|
sri |
the wiki entry might be based on a bug |
| 21:35 |
|
curt |
I worked through the example the other day without issue... on 0.999930 |
| 21:35 |
|
curt |
and it worked fine |
| 21:35 |
|
curt |
upgrade to 1.00 yesterday and can't make it work |
| 21:36 |
|
sri |
relative paths were handled wrong |
| 21:37 |
|
sri |
it was actually a rather huge bug that broke base tags |
| 21:37 |
|
curt |
so, is the example incorrect? |
| 21:37 |
|
sri |
ye |
| 21:38 |
|
curt |
is htt:///cities the correct syntax? |
| 21:38 |
|
sri |
no |
| 21:38 |
|
sri |
what are you trying to achieve? |
| 21:39 |
|
curt |
I'm just working the example |
| 21:40 |
|
* sri |
did never read it |
| 21:40 |
|
yko |
rendering <%= form_for '/example' %> at 'http://localhost/foo/bar' he want to get form with action 'http://localhost/example' i think |
| 21:40 |
|
curt |
yes |
| 21:41 |
|
curt |
that's the way it worked on 0.999930 |
| 21:41 |
|
yko |
->parse on non-empty Mojo::URL can behave little bit strange |
| 21:41 |
|
sri |
<%= form_for 'example' %>? |
| 21:41 |
|
sri |
yko: it has been aligned with html rules for urls |
| 21:42 |
|
curt |
I've try the following: <%= form_for '/example' %> |
| 21:42 |
|
curt |
<%= form_for 'example' %> |
| 21:42 |
|
curt |
<%= form_for '//example' %> |
| 21:43 |
|
curt |
all route to http://localhost/example/example |
| 21:43 |
|
sri |
ok now i get it |
| 21:43 |
|
sri |
you did in fact rely on broken behavior |
| 21:44 |
|
sri |
where it would generate an absolute path starting with a / |
| 21:44 |
|
sri |
which made apps non portable |
| 21:44 |
|
sri |
<%= base_tag %> |
| 21:44 |
|
sri |
add that to your layout |
| 21:44 |
|
sri |
or head section |
| 21:44 |
|
yko |
or use route names to generate paths |
| 21:44 |
|
yko |
*urls |
| 21:45 |
|
sri |
doesn't work |
| 21:45 |
|
sri |
paths are all portable now |
| 21:45 |
|
sri |
as in relative |
| 21:45 |
|
sri |
the problem is if a path starts with / it is absolute |
| 21:45 |
|
yko |
oh... i was dreaming about that half year ago :) |
| 21:45 |
|
sri |
bound to the current authority |
| 21:46 |
|
sri |
thats non portable when your application is actually based in http://localhost/myapp for example |
| 21:46 |
|
yko |
ye. that was my problem which i solved with tricky stuff |
| 21:47 |
|
sri |
was a very nasty bug |
| 21:47 |
|
sri |
lucky we got it fixed before 1.0 |
| 21:47 |
|
yko |
ye |
| 21:48 |
|
sri |
curt: quick solution <%= base_tag %> in your head section |
| 21:51 |
|
sri |
form_for doesn't actually generate "http://localhost/example/example", thats why i got confused |
| 21:51 |
|
sri |
it generates just a relative "example" |
| 21:55 |
|
|
GitHub1 joined #mojo |
| 21:55 |
|
GitHub1 |
mojo: master Sebastian Riedel * c21e164 (3 files in 2 dirs): updated application generators with base tags - http://bit.ly/ekjFKT |
| 21:55 |
|
|
GitHub1 left #mojo |
| 21:58 |
|
sri |
gonna update the examples too to make it more obvious |
| 21:59 |
|
|
jwang joined #mojo |
| 21:59 |
|
|
achromic left #mojo |
| 22:00 |
|
curt |
thanks, that resolved the issue... |
| 22:00 |
|
curt |
once I started calling the default layout |
| 22:03 |
|
curt |
might want to add "% layout 'default';" to the new_form.html.ep example |
| 22:03 |
|
sri |
it's a wiki, feel free |
| 22:05 |
|
curt |
don't have a git account... it might be time |
| 22:05 |
|
sri |
indeed |
| 22:05 |
|
|
achromic joined #mojo |
| 22:10 |
|
|
kvorg left #mojo |
| 22:12 |
|
|
GitHub93 joined #mojo |
| 22:12 |
|
GitHub93 |
mojo: master Sebastian Riedel * 8f21bc2 (1 files in 1 dirs): pod cleanup - http://bit.ly/hKmeUB |
| 22:12 |
|
|
GitHub93 left #mojo |
| 22:12 |
|
sri |
ok, that should help in the future |
| 22:13 |
|
marcus |
hmm, I think it would be easy to extend MojoX::Validator to generate data for jquery validation |
| 22:14 |
|
|
GitHub32 joined #mojo |
| 22:14 |
|
GitHub32 |
mojo: master Sebastian Riedel * d929cf8 (1 files in 1 dirs): pod cleanup - http://bit.ly/h5Ubaw |
| 22:14 |
|
|
GitHub32 left #mojo |
| 22:31 |
|
sri |
marcus: you might not want to put too much effort into it |
| 22:31 |
|
sri |
considering html5 forms |
| 22:33 |
|
|
su-bzero1 joined #mojo |
| 22:38 |
|
marcus |
what about html5 forms? |
| 22:38 |
|
|
su-bzero_ left #mojo |
| 22:38 |
|
|
GitHub169 joined #mojo |
| 22:38 |
|
GitHub169 |
mojo: master Sebastian Riedel * 1af6d84 (1 files in 1 dirs): pod cleanup - http://bit.ly/fVvn03 |
| 22:38 |
|
|
GitHub169 left #mojo |
| 22:38 |
|
marcus |
sri: html5 forms seems to be pretty similar to previous versions? Just more types? |
| 22:38 |
|
sri |
built in validation |
| 22:39 |
|
marcus |
just for very simple cases |
| 22:39 |
|
sri |
see "pattern" attribute ;) |
| 22:41 |
|
sri |
http://www.alistapart.com/arti[…]ready-for-html-5/ |
| 22:42 |
|
|
seb_ joined #mojo |
| 22:43 |
|
marcus |
not supported by chrome yet :) |
| 22:44 |
|
sri |
excuses |
| 22:45 |
|
marcus |
anyways, I need this for the project I am doing this week |
| 22:45 |
|
marcus |
and I don't think jquery validation will be obsolete too soon |
| 22:45 |
|
marcus |
the server will still need to validate html5 forms too |
| 22:45 |
|
marcus |
t |
| 22:45 |
|
marcus |
this is a convenient way to have a single defintiion of validation rules |
| 22:45 |
|
marcus |
definition |
| 22:46 |
|
marcus |
can always write some simple js to use the metadata to add attributes to a html form later |
| 22:55 |
|
|
GitHub127 joined #mojo |
| 22:55 |
|
GitHub127 |
mojo: master Sebastian Riedel * 2ed39b9 (1 files in 1 dirs): fixed typo - http://bit.ly/gKktzM |
| 22:55 |
|
|
GitHub127 left #mojo |
| 22:56 |
|
perlrocks |
Twitter: "Mojolicious:: Lite in DBIx:: Skinny've used it - it filtered its programmers Isuzu http://bit.ly/eBUTfh Sunday" (ja) --wPerlnews http://twitter.com/wPerlnews/s[…]21701286609031168 |
| 22:56 |
|
|
verrens joined #mojo |
| 22:56 |
|
sri |
hmm, maybe i should encourage async a little less |
| 22:57 |
|
sri |
it's an optimization that comes with a cost |
| 22:57 |
|
marcus |
and the price is YOUR SOUL |
| 22:57 |
|
sri |
marcus: actually i'm not sure anymore your plugin should be async by default |
| 22:58 |
|
sri |
the price is the debug screen |
| 22:58 |
|
sri |
async always moves request handling out of the safety net |
| 22:59 |
|
|
kaare left #mojo |
| 22:59 |
|
marcus |
sri: yeah, it complicates things |
| 22:59 |
|
sri |
i guess plugins should treat it as an optimization as well |
| 22:59 |
|
marcus |
sri: can't you magically put it back into the safety net? :) |
| 22:59 |
|
sri |
and only go async when asked |
| 23:00 |
|
sri |
marcus: i wish :) |
| 23:00 |
|
marcus |
sri: I guess the plugin should still use callbacks tho |
| 23:00 |
|
sri |
ye |
| 23:00 |
|
marcus |
to have a coherent api |
| 23:00 |
|
sri |
you basically just need something like my $client = $async ? $client->async : $client; |
| 23:00 |
|
marcus |
mm |
| 23:01 |
|
marcus |
I guess I shouldn't be calling render_later if it's not async |
| 23:01 |
|
sri |
oh right |
| 23:02 |
|
marcus |
$c->render_later if $args{async}; |
| 23:02 |
|
* sri |
nods |
| 23:03 |
|
marcus |
I guess the async version isn't really faster |
| 23:03 |
|
marcus |
it just frees up the daemon from blocking |
| 23:03 |
|
sri |
exactly |
| 23:04 |
|
sri |
actually non blocking is usually a little slower, you can just do something else while waiting |
| 23:06 |
|
marcus |
weird |
| 23:06 |
|
marcus |
automatic rendering triggered even on non-asyn |
| 23:06 |
|
marcus |
async |
| 23:06 |
|
marcus |
without render_later |
| 23:07 |
|
marcus |
ignore me |
| 23:07 |
|
* sri |
ignores marcus |
| 23:07 |
|
marcus |
it hangs instead |
| 23:07 |
|
marcus |
of course |
| 23:07 |
|
marcus |
the test can't work without async :) |
| 23:07 |
|
sri |
Oo |
| 23:08 |
|
sri |
sounds like you did go async |
| 23:08 |
|
marcus |
since it's hitting localhost in a single-thread |
| 23:08 |
|
sri |
oooh |
| 23:08 |
|
marcus |
guess the test gets async on then |
| 23:08 |
|
sri |
actually that shouldn't be a problem |
| 23:09 |
|
sri |
https://github.com/kraih/mojo/[…]s/lite_app.t#L384 |
| 23:10 |
|
sri |
this one does a "blocking" subrequest too |
| 23:10 |
|
sri |
you just give it a path |
| 23:10 |
|
sri |
it will fire up a second daemon in another ioloop |
| 23:10 |
|
marcus |
yeah, but I'm feeding the client the full url |
| 23:11 |
|
marcus |
token_url => Mojo::URL->new("http://localhost:$port/fake_token"), |
| 23:11 |
|
sri |
don't do that for the blocking test :) |
| 23:11 |
|
sri |
/fake_token to test blocking |
| 23:11 |
|
* marcus |
adds blocking.t |
| 23:16 |
|
* marcus |
gets fails |
| 23:16 |
|
* marcus |
cries |
| 23:17 |
|
marcus |
Mon Jan 3 00:17:07 2011 error Mojo::Client:763 [71960]: Couldn't connect. |
| 23:17 |
|
marcus |
guess it doesn't work if the url is a Mojo::URL object? |
| 23:18 |
|
sri |
don't think it matters |
| 23:21 |
|
|
verrens left #mojo |
| 23:30 |
|
marcus |
sri: "/$fb_url" |
| 23:30 |
|
marcus |
if I do it like this it works |
| 23:30 |
|
sri |
weird |
| 23:30 |
|
sri |
ooooh |
| 23:31 |
|
sri |
relative url of course |
| 23:31 |
|
marcus |
sri: yeah, but I'm not sure the relative handling is sane |
| 23:31 |
|
marcus |
I do Mojo::URL->new("/foo") |
| 23:31 |
|
marcus |
and it stringifies to "foo" |
| 23:32 |
|
sri |
feel free to improve it |
| 23:32 |
|
marcus |
perl -MMojo::URL -le'print Mojo::URL->new("/foo").""' |
| 23:32 |
|
marcus |
I think I'll go to bed instead |
| 23:32 |
|
sri |
:O |
| 23:32 |
|
sri |
slacker |
| 23:33 |
|
sri |
in the old days you have stayed up all night! |
| 23:33 |
|
marcus |
It's already 12:30 |
| 23:33 |
|
marcus |
I'm pretty sure I didn't |
| 23:34 |
|
marcus |
I've been married the whole time we've been hacking together :p |
| 23:34 |
|
sri |
bad memory too! |
| 23:34 |
|
sri |
nn pops |
| 23:36 |
|
* marcus |
turns on the booby traps on his lawn and goes to bed. |
| 23:36 |
|
|
kvorg joined #mojo |
| 23:44 |
|
sri |
marcus: actually there might be a better solution |
