00:27
<_jud>
Anyone know why a connection to the travis Redis server would be timing out?
00:27
<parndt>
any number of reasons
00:27
<parndt>
antares_ ^
00:28
<_jud>
any number that I could control? Just pushing my first branch to travis
00:56
<GitHub192>
[travis-ci.github.com] randym pushed 1 new commit to next_release: http://git.io/HPyG5Q
00:56
<GitHub192>
[travis-ci.github.com/next_release] beginnings of next release blog - on to you @josh-k - Randy Morgan
02:31
<SubStack>
if I push a commit will that kill a pending build that's hung?
02:31
<SubStack>
or should I wait until it times out
02:32
<parndt>
it won't kill it
02:32
<parndt>
it'll just eventually time out
02:32
<SubStack>
if I push and there is a hung test will it run?
02:33
<SubStack>
oh wait, there it goes
02:33
<SubStack>
sweet, all passed
02:33
<SubStack>
it hung for like 3 minutes on the `npm install` part before doing anything >_<
02:34
<GitHub127>
[travis-ci] ifesdjeen pushed 1 new commit to master: http://git.io/0824jQ
02:34
<GitHub127>
[travis-ci/master] Removing token information from payload. - Alex P
02:40
<travis-ci>
[travis-ci] travis-ci/travis-ci#1570 (master - aeac73c : Alex P): The build passed.
02:40
<travis-ci>
[travis-ci] Change view : https://github.com/travis-ci/travis-ci/compare/7b79eb0...aeac73c
02:40
<travis-ci>
[travis-ci] Build details : http://travis-ci.org/travis-ci/travis-ci/builds/749205
02:41
<SubStack>
so yay, that project finally has a green badge
02:42
<SubStack>
is there a way to link to all the projects a particular user has on travis?
02:45
<parndt>
I'm not sure that there is
02:45
<parndt>
at least I haven't seen a way
04:09
renan_saddam joined
04:12
<GitHub49>
[travis-ci] mattmatt pushed 1 new commit to master: http://git.io/M0kawg
04:12
<GitHub49>
[travis-ci/master] Merge branch 'master' into logging - Mathias Meyer
04:15
<GitHub54>
[travis-ci] mattmatt pushed 1 new commit to master: http://git.io/4T3Z-w
04:15
<GitHub54>
[travis-ci/master] Update gitignore - Mathias Meyer
04:18
<randym>
josh-k: ping
04:18
<travis-ci>
[travis-ci] travis-ci/travis-ci#1571 (master - bd17ef0 : Mathias Meyer): The build passed.
04:18
<travis-ci>
[travis-ci] Change view : https://github.com/travis-ci/travis-ci/compare/aeac73c...bd17ef0
04:18
<travis-ci>
[travis-ci] Build details : http://travis-ci.org/travis-ci/travis-ci/builds/749607
04:18
<josh-k>
randym: one sec
04:19
<randym>
no worries - heading out the door - Ill catch you later re: blog post
04:21
<travis-ci>
[travis-ci] travis-ci/travis-ci#1573 (production - 5ddce23 : Mathias Meyer): The build passed.
04:21
<travis-ci>
[travis-ci] Change view : https://github.com/travis-ci/travis-ci/compare/7b79eb0...5ddce23
04:21
<travis-ci>
[travis-ci] Build details : http://travis-ci.org/travis-ci/travis-ci/builds/749621
04:21
<travis-ci>
[travis-ci] travis-ci/travis-ci#1572 (master - 5ddce23 : Mathias Meyer): The build passed.
04:21
<travis-ci>
[travis-ci] Change view : https://github.com/travis-ci/travis-ci/compare/bd17ef0...5ddce23
04:21
<travis-ci>
[travis-ci] Build details : http://travis-ci.org/travis-ci/travis-ci/builds/749618
04:48
<josh-k>
randym: ping?
04:51
<randym>
hey josh-k ^ just about to get on a train. Can you have a look at the 'three_new_features' blog post in the next_release branch of travis-ci.github.com please?
04:51
<randym>
Ill be back online later
04:52
<josh-k>
randym: <3<3<3
04:52
<josh-k>
i should be online all day, ping me when you are back
04:53
<parndt>
ugh I forgot to hook up travis before pushing
04:54
<parndt>
josh-k: is there any way to trigger a build on a branch other than master without pushing again?
04:54
<josh-k>
parndt: nope sorry
04:54
<parndt>
alright :)
04:54
<parndt>
thanks buddy
04:54
<parndt>
I'll have to justify one more commit somehow
04:56
<josh-k>
parndt: hahaha
04:57
<parndt>
josh-k: ka-ching! https://github.com/parndt/xrono/commit/bda32f36739118c1171ac83eba7d46fdf6d8936a
05:08
AutomatedTester joined
05:47
<GitHub160>
[travis-ci] mattmatt pushed 4 new commits to master: http://git.io/2cbVhw
05:47
<GitHub160>
[travis-ci/master] Rescue routing errors. - Mathias Meyer
05:47
<GitHub160>
[travis-ci/master] Use a fallback route for routing errors instead. - Mathias Meyer
05:47
<GitHub160>
[travis-ci/master] Fix route_not_found action. - Mathias Meyer
05:47
benediktdeicke joined
05:52
<antares_>
josh-k: hey
05:53
<josh-k>
antares_: hey :)
05:53
<antares_>
josh-k: I found https://github.com/mitchellh/vagrant/issues/516 last night
05:53
<roidrage>
cleaning up logging, one commit and deploy at a time ;)
05:54
<travis-ci>
[travis-ci] travis-ci/travis-ci#1574 (master - 7a25ca6 : Mathias Meyer): The build passed.
05:54
<travis-ci>
[travis-ci] Change view : https://github.com/travis-ci/travis-ci/compare/5ddce23...7a25ca6
05:54
<travis-ci>
[travis-ci] Build details : http://travis-ci.org/travis-ci/travis-ci/builds/749999
05:54
<antares_>
josh-k: I am afraid we have to try to find a workaround or not make it a particularly big launch
05:54
<antares_>
pip opens so many connections it happens every 3rd time or so
05:54
<travis-ci>
[travis-ci] travis-ci/travis-ci#1575 (production - 7a25ca6 : Mathias Meyer): The build passed.
05:54
<travis-ci>
[travis-ci] Change view : https://github.com/travis-ci/travis-ci/compare/5ddce23...7a25ca6
05:54
<travis-ci>
[travis-ci] Build details : http://travis-ci.org/travis-ci/travis-ci/builds/750002
05:55
<josh-k>
antares_: well, we can't dictate if it is a big launch or not, either we find a workaround or don't launch
05:56
<antares_>
the only workaround that seems to help is to change NIC virtualbox uses. The only example we have is for vagrant 0.9. We will need to update the worker to it, I already updated sous-chef
05:56
<antares_>
josh-k: yeah that's true
05:58
<antares_>
josh-k: I already updated sous-chef to 0.9: https://github.com/michaelklishin/sous-chef/blob/03e42f05721085e6ad3ceb7cb200955d51967a87/Vagrantfile.sample
05:58
<josh-k>
antares_: i have updated travis-boxes to use vagrant 0.9
05:58
<josh-k>
just need to push my changes
05:59
<josh-k>
can you point me to a log which shows this error on Travis?
05:59
<antares_>
can you update travis-worker? I will recreate VMs on php1 and php2
06:01
<josh-k>
antares_: i will update the worker shortly
06:03
<josh-k>
antares_: i have mitchell on IM, not online now of course, so we can talk to him about this further
06:04
<randym>
josh-k: ping when you have a sec.
06:06
<antares_>
josh-k: right. So far I see that switching NIC to what I have in sous-chef now helps (fingers crossed).
06:06
<josh-k>
antares_: has this error occured only during provisioning or during test runs?
06:07
<josh-k>
randym: sure, will do :)
06:07
<GitHub18>
[travis-cookbooks] michaelklishin pushed 1 new commit to master: http://git.io/C27ULw
06:07
<GitHub18>
[travis-cookbooks/master] Looks like NIC changes in Vagrantfile are sufficient, comment out this extra workaround - Michael S. Klishin
06:07
<antares_>
josh-k: sometimes during test runs, too
06:08
<GitHub187>
[travis-boxes] joshk pushed 2 new commits to master: http://git.io/62Hf0g
06:08
<GitHub187>
[travis-boxes/master] change to vagrant 0.9 - Josh Kalderimis
06:08
<GitHub187>
[travis-boxes/master] use vagrant programmatically - Josh Kalderimis
06:08
<antares_>
josh-k: mostly for Python but this also explains dropped SSH connection soon after deploying new VM images (when all workers are installing dependencies at once after coming back online)
06:08
<josh-k>
antares_: do you have a link to one of these failures?
06:09
<antares_>
I don't but they are completely random as described in the ticket
06:09
<antares_>
josh-k: locally with sous-chef, this seems to help: https://github.com/michaelklishin/sous-chef/blob/03e42f05721085e6ad3ceb7cb200955d51967a87/Vagrantfile.sample#L20
06:10
<josh-k>
antares_: sure, but the ticket is not a travis ticket, so i was wondering if we have an example of it happening for tests?
06:10
<antares_>
josh-k: it happened 100% of the time for me with sous-chef
06:10
<antares_>
josh-k: when I tried to provision python::multi with several packages we preinstall
06:11
<antares_>
it was 5 pythons * 3 packages = 15 packages installed with --use-mirrors in a row
06:11
<josh-k>
ok, sure, but i asked if it happened with tests and you said yes, so do we have any logs?
06:11
<antares_>
josh-k: I don't have any logs immediately but often when we deploy new images a few initial tests drop ssh connections, then it is all fine again
06:12
<GitHub103>
[travis-boxes] joshk pushed 1 new commit to master: http://git.io/w18Keg
06:12
<GitHub103>
[travis-boxes/master] test against jruby 1.8 - Josh Kalderimis
06:12
<josh-k>
antares_: ok, so this isn't an issue we need to hold back the launch for
06:12
<antares_>
josh-k: it is a significantly bigger deal for python
06:12
<josh-k>
sure, but we seem to have it under control, otherwise we would see stupid failures
06:13
<antares_>
josh-k: pip seems to be way more aggressive in opening network connections than bundler, maven or anything else
06:13
<antares_>
josh-k: well, lets try deploying this NIC change then ask people around here trigger a dozen of python builds
06:13
<antares_>
I hope so
06:13
<antares_>
but nothing is worse than having 300 projects join in a day and seeing this issue 50% of the time
06:14
<josh-k>
sorry, i really don't understand the problem. python images are deployed and we have run many many tests on them without seeing this issue
06:14
<antares_>
josh-k: WITH SEEING THEM
06:14
<josh-k>
we won't get 300, we will get 50-70
06:14
<antares_>
I just did not understand why exactly
06:14
<antares_>
but in that thread I see half of the people mentioning pip
06:14
<antares_>
and it only happens locally for me when something is installing packages via pip
06:15
<antares_>
maybe jezdez can tell us how we can limit # of connections pip opens at once
06:15
<josh-k>
antares_: they all talk about provisioning
06:15
<antares_>
josh-k: at the bottom you see 2 people figuring out the cause
06:15
<josh-k>
anyhow, this issue should not hold up launch
06:15
<josh-k>
i know, i saw that
06:16
<antares_>
josh-k: virtualbox NIC using NAT starts dropping connections when it has way too many outgoing connections
06:16
<antares_>
and ssh is usually one of the first connections it has (inbound)
06:16
<antares_>
josh-k: we should still deploy 0.9 and NIC update to php1 and php2
06:16
<josh-k>
hmmmmmmmmmmmmmmmm
06:16
<antares_>
then we will see what mitchell can suggest us
06:17
<josh-k>
well, i think we are blowing this issue up, i would prefer to get stuff out the door which works instead of trying to get this all done and ready for 5 or 6 hours time
06:17
<josh-k>
i don't doubt this is an issue
06:17
<josh-k>
but is it something which we should rush to fix and potentially delay us more?
06:18
<antares_>
josh-k: if this happens even 1/4 of the time, there will be a lot of pissed off people and snarky comments
06:19
<josh-k>
antares_: we have been running for a year and we have not seen this during test runs
06:19
<antares_>
josh-k: we have
06:19
<antares_>
with python, especially often
06:19
<antares_>
every once in a while
06:19
<josh-k>
what test runs?
06:20
<antares_>
mostly Ruby, with huge Gemfiles
06:20
<josh-k>
i have NEVER seen this
06:20
<antares_>
with pip you don't even need a lot of dependencies for this to happen
06:20
<antares_>
I see this every week
06:20
<josh-k>
then lets reproduce it and then worry
06:20
<josh-k>
i have never seen this
06:20
<antares_>
josh-k: well, ok
06:20
<antares_>
lets just launch this as is
06:20
<antares_>
but I warned everyone about this
06:21
<antares_>
at least there seems to be an easy workaround
06:21
<antares_>
that can be deployed in ~ 1 hour instead of days
06:21
<josh-k>
sorry but don't pull that 'i warned you' crap, if you are going to say that so later you can say 'told you so' then we won't launch
06:22
<mmalecki>
hi. not sure what's going on, but can I help somehow?
06:22
<josh-k>
if you can show me a log where this happens during a test run then we can worry and fix, but getting travis worker updated and rebuilding vms and trying the NIC changes within 4-5 hours is a little of a stretch
06:23
<josh-k>
antares_: so how about this, you decide if we delay by a day, and decide now, because if we rush to fix this and then don't have time to test it then that will look even worse
06:23
<antares_>
let me find a log
06:24
<GitHub196>
[travis-ci] svenfuchs pushed 1 new commit to master: http://git.io/7CjHTw
06:24
<GitHub196>
[travis-ci/master] Merge branch 'master' of github.com:travis-ci/travis-ci - Sven Fuchs
06:24
<antares_>
josh-k: http://travis-ci.org/#!/kennethreitz/requests/jobs/745980
06:24
<antares_>
josh-k: requests make lots of HTTP requests but even before that I have seen issues during pip install runs
06:25
<josh-k>
this is after pip install
06:25
<josh-k>
so again, you decide because i don't want you saying 'oh i told you so'
06:25
<josh-k>
if we delay this then we delay it by 24-48 hours so we can test it
06:25
<antares_>
http://travis-ci.org/#!/kennethreitz/requests/jobs/745975 has the same issue
06:27
beawesomeinstead joined
06:27
beawesomeinstead joined
06:27
beawesomeinstead joined
06:27
<antares_>
josh-k: I won't say "I told you so". After reviewing requests build history it seems to only affect python 3.2.
06:28
<antares_>
josh-k: we can ask kennethreitz about this. Other Python projects we have seem to be fine, we just tried kennethreitz/requests many more times than any other repo.
06:28
<josh-k>
antares_: i am going to make another coffee, you decide what we do, either go live today or delay by a day so we have time to talk to Mitchell
06:29
<antares_>
josh-k: so I think there may be selection bias. I think we can launch today.
06:30
<GitHub59>
[travis-boxes] joshk pushed 1 new commit to master: http://git.io/L9R1Ag
06:30
<GitHub59>
[travis-boxes/master] Revert "test against jruby 1.8" - Josh Kalderimis
06:30
TheDeadSerious joined
06:31
<travis-ci>
[travis-ci] travis-ci/travis-ci#1576 (master - 48f3afc : Sven Fuchs): The build passed.
06:31
<travis-ci>
[travis-ci] Change view : https://github.com/travis-ci/travis-ci/compare/7a25ca6...48f3afc
06:31
<travis-ci>
[travis-ci] Build details : http://travis-ci.org/travis-ci/travis-ci/builds/750137
06:31
<josh-k>
travis-boxes now uses vagrant 0.9 and vagrant programmatically
06:31
<josh-k>
it should work without issue on jruby
06:32
<josh-k>
if you want to try your NIC changes then you can
06:32
<svenfuchs>
that sounds cool
06:32
<josh-k>
i am going to update travis-worker so it works with jruby better
06:32
<antares_>
svenfuchs: hi. Have you seen our discussion about vagrant issue #516?
06:33
<svenfuchs>
not really :)
06:33
<antares_>
svenfuchs: apparently many SSH connection failures are not due to net/ssh or ubuntu's sshd issues. They happen when stuff in VMs opens too many outgoing network connections at once. Virtualbox virtual network card using NAT then starts dropping connections (intentionally or not).
06:34
<travis-ci>
[travis-ci] travis-ci/travis-ci#1577 (production - 48f3afc : Sven Fuchs): The build passed.
06:34
<travis-ci>
[travis-ci] Change view : https://github.com/travis-ci/travis-ci/compare/7a25ca6...48f3afc
06:34
<travis-ci>
[travis-ci] Build details : http://travis-ci.org/travis-ci/travis-ci/builds/750141
06:34
<antares_>
svenfuchs: there is a fairly quick workaround that requires upgrading to vagrant 0.9. So hopefully this will be less of an issue in the future.
06:34
<svenfuchs>
wow, that sounds wicked
06:36
<svenfuchs>
sooo, will the ppp release be delayed because of that?
06:36
<josh-k>
i personally think we should continue with the release
06:37
<antares_>
josh-k: we will
06:37
<josh-k>
if we delay it then we delay it by a day
06:37
<antares_>
it may be the exact project we did most of python testing on, it opens a lot of network connections in its test suite because it is an HTTP client
06:40
<josh-k>
i am updating travis-worker to vagrant 0.9
06:42
<antares_>
When I am trying to build a box with travis-boxes on 0.9, it complains that base MAC address for eth0/NAT must be set. Not sure why yet, we may need to rebuild base boxes, too.
06:42
<josh-k>
try a second time
06:42
<josh-k>
i had that issue once, i think i need to add vagrant.boxes.reload!
06:43
<antares_>
I found this so far: https://github.com/mitchellh/vagrant/issues/564
06:44
<josh-k>
hmmm, says he fixed it
06:44
<cogocogo>
Is there any document describing how to get the spec stuff up and running for travis-core? I have implemented a small feature but can't execute the spec for it. ALERT: Total RoR n00b.
06:45
<josh-k>
cogocogo: ohhhh what is the feature? :)
06:45
<cogocogo>
josh-k: Option for letting the irc-bot send NOTICE instead of PRIVMSG.
06:45
<josh-k>
interesting, cool!
06:46
<cogocogo>
josh-k: I have implemented the feature, but can't get the spec stuff working, so I have no idea of my implementation works as expected. :p
06:46
<josh-k>
cogocogo: you can be lazy and just add it to travis and have travis do the testing?
06:47
<cogocogo>
Well, of course ... but that's a dick move, and I don't want to be a dick. :p
06:47
<cogocogo>
But sure. I can push the changes and let some of you guys review it anyways. :)
06:47
<josh-k>
cogocogo: it isn't a dick move, it is what we are for :)
06:49
argent-smith joined
06:51
<cogocogo>
josh-k: You one of the guys who can look at the feature and decide whether it needs fixing or not before it can be merged?
06:51
<josh-k>
cogocogo: i am one of those guys, yes :)
06:51
<parndt>
one of the most handsome, even
06:51
<antares_>
cogocogo: can you explain why that feature is necessary at all?
06:52
<antares_>
our notifications code already has many special cases and conditions
06:53
<cogocogo>
antares_: It's annoying when bots do PRIVMSG instead of NOTICE since it causes most clients to inform me that a user as said something on the channel. And it's easier for users who don't really need to see all the message to filter them out.
06:53
<cogocogo>
The jenkins bot has some options for it as well.
06:53
<antares_>
cogocogo: maybe NOTICE then can be a new default? what does github use, for example?
06:54
<cogocogo>
hmm ... not sure.
06:55
<cogocogo>
There is an option for notices
06:55
<antares_>
so github notifications also has this option?
06:56
<cogocogo>
Yup. At least in the admin GUI :p
06:56
<GitHub86>
[travis-core] mattmatt created rails-3.2 (+1 new commit): http://git.io/Oc5eCw
06:56
<GitHub86>
[travis-core/rails-3.2] Start Rails 3.2 migration. - Mathias Meyer
06:56
<antares_>
if so I guess we should have it, too
06:58
<cogocogo>
antares_: josh-k: https://github.com/christeredvartsen/travis-core/commit/4e7a7273114d805f533cb50fc93bd18e3d8b1389 ... Keep in mind that I'm rather new to ruby, so I don't really know if this is the best way to solve the problem.
06:59
<cogocogo>
If it looks OK I can fire off a PR, and you might be able to fix the spec stuff.
07:00
<josh-k>
cogocogo: hmmm, sorry, the implementation is not quite right
07:00
<cogocogo>
hehe, ok. :)
07:00
<josh-k>
cogocogo: i would suggest finding the irc specs and start there
07:01
<josh-k>
i would help more but i need to make some changes to the worker right now
07:01
<cogocogo>
josh-k: Thats what I tried, but I can't figure out how to run the spec file. :/
07:01
<cogocogo>
josh-k: I'll try some more and update the branch when I have it running. :)
07:02
<travis-ci>
[travis-ci] travis-ci/travis-core#271 (rails-3.2 - 9757c5a : Mathias Meyer): The build failed.
07:02
<travis-ci>
[travis-ci] Change view : https://github.com/travis-ci/travis-core/commit/9757c5a
07:02
<travis-ci>
[travis-ci] Build details : http://travis-ci.org/travis-ci/travis-core/builds/750204
07:02
<antares_>
cogocogo: see .travis.yml for how to run tests
07:04
<cogocogo>
antares_: OK, thanks. I'll start there.
07:04
<josh-k>
cogocogo: sorry i can't help more now
07:05
<parndt>
goodnight <3
07:05
<josh-k>
parndt: slaap lekker!
07:05
<cogocogo>
josh-k: NP. It's better for learning anyways. :)
07:06
<antares_>
parndt: good night!
08:05
mikespokefire joined
08:52
<josh-k>
antares_: i am also getting that "* Base MAC address for eth0/NAT must be set. Contact box maintainer for more information." error, i need to wait for Mitchell
08:53
<antares_>
josh-k: ack
08:54
<josh-k>
jezdez: ping?
08:59
<GitHub135>
[travis-ci] svenfuchs pushed 1 new commit to production: http://git.io/MRoe-w
08:59
<GitHub135>
[travis-ci/production] add codeminer as a silver sponsor - Sven Fuchs
09:04
<travis-ci>
[travis-ci] travis-ci/travis-ci#1578 (production - edd24d1 : Sven Fuchs): The build passed.
09:04
<travis-ci>
[travis-ci] Change view : https://github.com/travis-ci/travis-ci/compare/48f3afc...edd24d1
09:04
<travis-ci>
[travis-ci] Build details : http://travis-ci.org/travis-ci/travis-ci/builds/750762
09:13
geekoncoffee joined
09:28
<svenfuchs>
so, where's the release? :)
09:28
<GitHub126>
[travis-core] mattmatt pushed 1 new commit to rails-3.2: http://git.io/FEBZ7w
09:28
<GitHub126>
[travis-core/rails-3.2] Bump libraries for JRuby. - Mathias Meyer
09:29
<antares_>
roidrage: dbcleaner 0.7.1 used to break JRuby
09:29
<josh-k>
svenfuchs: 5pm
09:29
<josh-k>
svenfuchs: sorry, forgot to tell you it was moved a little, i told roidrage and antares_
09:29
<roidrage>
antares_: now 0.7.0 breaks jruby
09:29
<antares_>
roidrage: the entire test run failed because of obscure exceptions
09:29
<roidrage>
it doesn't anymore
09:29
<antares_>
roidrage: ah, maybe it was 0.7.0
09:29
<roidrage>
it failed with rails 3.2 and database_cleaner 0.7.0 actually
09:29
<antares_>
I remember I reverted it before
09:30
<antares_>
roidrage: so many failures, can't even keep them straight…
09:30
<roidrage>
yeah, i'm down to one error on jruby lcoally
09:30
<antares_>
josh-k, roidrage, svenfuchs, khaase: just to clarify, we will announce python and perl release today at 17 CET
09:31
<roidrage>
despite the connection woes?
09:32
<antares_>
roidrage: we figured it doesn't affect most of projects. The one we used for testing is an HTTP client so it issues a ton of requests.
09:32
<roidrage>
we had a chat with jezdez over lunch, he might have a fix for that ;)
09:32
<josh-k>
roidrage: i tried pinging him but he didn't respond
09:32
<josh-k>
jezdez: HOLLAAAA!
09:32
<antares_>
it is definitely something we will try to fix in upcoming days (there is a workaround, maybe pip can be configured, too) but most projects probably won't be affected
09:33
<dstufft>
he might be working
09:33
<dstufft>
he's somewhere in Euroland, Germany I think
09:33
<josh-k>
dstufft: he is based in Berlin
09:33
<dstufft>
yea that's it
09:33
<josh-k>
dstufft: just down the road from the Travis office
09:33
<* dstufft>
just woke up
09:33
<roidrage>
he's actually just a five minute walk from the office ;)
09:33
<josh-k>
dstufft: we know him personally
09:33
<dstufft>
oh how about
09:34
<dstufft>
meanwhile i'm stuck in Philadelphia D:
09:36
<roidrage>
isn't it always sunny there?
09:36
<josh-k>
roidrage: i was about to say that!
09:42
<roidrage>
wheee, builds are green for travis-core on rails 3.2
09:42
<travis-ci>
[travis-ci] travis-ci/travis-core#272 (rails-3.2 - ffad42b : Mathias Meyer): The build was fixed.
09:42
<travis-ci>
[travis-ci] Change view : https://github.com/travis-ci/travis-core/compare/9757c5a...ffad42b
09:42
<travis-ci>
[travis-ci] Build details : http://travis-ci.org/travis-ci/travis-core/builds/750865
09:42
<khaase>
antares_: cool
09:44
<nateabele>
dstufft: There are worse places to be than Philly, and much better places aren't far away.
09:44
<dstufft>
nateabele: don't mind me, I just hate where I live, and i'm stuck here for another 7 years or so
09:45
<nateabele>
dstufft: Maybe we should have a beer about that sometime, because I think I'm stuck in pretty much the same boat. ;-)
09:45
<dstufft>
nateabele: ah you're in philly too? I'm actually in norristown but most the world doesn't know what that is ;P
09:46
<nateabele>
I know that some part of my brain knows where it is, but I'm gonna have to Google Maps it anyway.
09:46
<dstufft>
king of prussia is next door to me
09:46
<nateabele>
Ah, okay, yeah that looks familiar.
09:47
<nateabele>
I'm in Ardmore, so almost due south of you.
09:47
<nateabele>
And yeah, it sucks. I moved here from Brooklyn about 5 months ago. Talk about your reversal of fortunte...
09:47
<nateabele>
* fortune
09:49
<dstufft>
I moved here from central PA
09:49
<dstufft>
like 6 years ago
09:49
<antares_>
josh-k: based on anecdotal evidence, the rails2 => ruby4 move makes a noticeable difference so far
09:50
<josh-k>
yep, it was a good move
09:50
<antares_>
josh-k: I see 22-24 parallel builds on ruby workers pretty often and only saw a long Rails queue once so far
09:50
<nateabele>
dstufft: Well that's still a step up. :-)
09:50
<josh-k>
when all ruby workers are busy, how is the hub holding up with processing the messages?
09:52
<antares_>
josh-k: as usual
09:52
<antares_>
pretty much always no backlog
09:52
<antares_>
when there is, it is hardly higher than 40-50 messages
09:53
<antares_>
I can't say the average msg/s rate went up by any significant amount
09:53
<antares_>
it is still far from 77 msg/s peak I saw once
09:54
<antares_>
also, I don't remember last time I had to do any maintenance for Hub
09:54
<antares_>
I think it was back in mid-January
09:54
<antares_>
and we have deployed updates since then
09:55
<antares_>
so, hub seems to be doing well
09:55
<josh-k>
we still need to think of a way to run multiple hubs and not have them fight over build logs (eg. only one hub should deal with one build log so the messages are processed in order)
09:56
<antares_>
one server-named queue + consumer per build, yes
09:56
<antares_>
well, this version should last us for several months
09:56
<antares_>
then we will see
09:57
<antares_>
I hope after this launch and VM images service I can focus on messaging parts
10:09
AutomatedTester joined
10:13
<GitHub124>
[travis-ci.github.com] michaelklishin pushed 1 new commit to master: http://git.io/OtDjEg
10:13
<GitHub124>
[travis-ci.github.com/master] Merge branch 'master' of github.com:travis-ci/travis-ci.github.com - Michael S. Klishin
10:14
<GitHub55>
[travis-support] mattmatt created rails-3.2 (+1 new commit): http://git.io/qjigVQ
10:14
<GitHub55>
[travis-support/rails-3.2] Comment out routing patches. - Mathias Meyer
10:14
<antares_>
please RT: https://twitter.com/travisci/status/174150415762472960
10:15
<josh-k>
http://news.ycombinator.com/newest and upvote :)
10:19
<dstufft>
antares_: if it matters the genesis video isn't available in america
10:19
<dstufft>
http://d.stufft.io/2p0M3R0T3M3A141K1X2H
10:19
<josh-k>
dstufft: REALLY??
10:19
<josh-k>
dstufft: can you find me a version of the vid which is available?
10:19
<josh-k>
we can update the links
10:20
<GitHub18>
[travis-support] svenfuchs pushed 1 new commit to rails-3.2: http://git.io/Dw0NNw
10:20
<GitHub18>
[travis-support/rails-3.2] remove rails routing monkey patch for devise, protect rspec monkey patch with gem-patching - Sven Fuchs
10:20
<dstufft>
http://
www.youtube.com/watch?v=1pkVLqSaahk this ones seems o be avail, assuming you wer elinking to the music video for it
10:21
<josh-k>
yep, antares_ can you please update the link?
10:21
<antares_>
give me a second
10:21
<dstufft>
assuming it's available in non america (the one I linked)
10:23
<josh-k>
it worked for me
10:23
<antares_>
dstufft: it is
10:23
<GitHub86>
[travis-ci.github.com] michaelklishin pushed 2 new commits to master: http://git.io/SVCYHg
10:23
<GitHub86>
[travis-ci.github.com/master] Update Land Of Confusion link to make sure US residents can watch it - Michael S. Klishin
10:23
<GitHub86>
[travis-ci.github.com/master] Regenerate - Michael S. Klishin
10:24
<dstufft>
regional blocking FTL
10:37
<svenfuchs>
whoooopooopoop
10:37
<svenfuchs>
https://skitch.com/svenfuchs/8frpe/travis-ci-announcing-python-and-perl-project-support-on-travis-ci
10:37
<svenfuchs>
antares_: do you think we can remove "project" from the headline?
10:37
<svenfuchs>
i know it's not really that important. but stuff like this makes my eyes bleed :)
10:38
<geekoncoffee>
so make your font smaller ;) j/k
10:38
<AutomatedTester>
antares_: wanna read before I press publish?
10:39
<antares_>
svenfuchs: sorry, remove what?
10:39
<antares_>
svenfuchs: we can, but I am not sure I understand what to remove
10:39
<antares_>
svenfuchs: also, this will break the URL :(
10:39
<antares_>
and thus our tweet link
10:39
<svenfuchs>
oooh, we can't keep the url?
10:39
<svenfuchs>
ok, forget about it then
10:39
<antares_>
let me see but I am afraid not
10:40
<GitHub89>
[travis-support] mattmatt pushed 1 new commit to rails-3.2: http://git.io/Dytb3A
10:40
<GitHub89>
[travis-support/rails-3.2] Require the gem_patching library. - Mathias Meyer
10:41
<GitHub48>
[travis-ci.github.com] michaelklishin pushed 1 new commit to master: http://git.io/ZL23hA
10:41
<GitHub48>
[travis-ci.github.com/master] Minor title correction - Michael S. Klishin
10:41
<antares_>
svenfuchs: it turns out we can, fixed
10:41
<svenfuchs>
fantastic, thanks :D
10:42
<antares_>
why doesn't it show up on the site :/
10:42
<josh-k>
kennethreitz: jezdez: and the announcement is live :)
10:42
<kennethreitz>
josh-k: upvoted :)
10:43
<antares_>
svenfuchs: GH pages are being slow but I updated the title and URL did not change (because it is calculated from the file name)
10:44
<GitHub10>
[travis-support] mattmatt pushed 1 new commit to rails-3.2: http://git.io/l0Rl_A
10:44
<GitHub10>
[travis-support/rails-3.2] Derp. - Mathias Meyer
10:45
<svenfuchs>
antares_: i see the change. perfect :)
10:46
<GitHub43>
[travis-support] mattmatt pushed 1 new commit to rails-3.2: http://git.io/rvROSg
10:46
<GitHub43>
[travis-support/rails-3.2] Ditch gem-patching for now. - Mathias Meyer
10:55
<GitHub181>
[travis-hub] mattmatt created rails-3.2 (+2 new commits): http://git.io/7VPCWA
10:55
<GitHub181>
[travis-hub/rails-3.2] Add .rbenv-version to gitignore - Mathias Meyer
10:55
<GitHub181>
[travis-hub/rails-3.2] Bump travis-core/travis-support - Mathias Meyer
10:58
<travis-ci>
[travis-ci] travis-ci/travis-hub#134 (rails-3.2 - bf8ff56 : Mathias Meyer): The build passed.
10:58
<travis-ci>
[travis-ci] Change view : https://github.com/travis-ci/travis-hub/compare/c656059^...bf8ff56
10:58
<travis-ci>
[travis-ci] Build details : http://travis-ci.org/travis-ci/travis-hub/builds/751367
10:59
<antares_>
josh-k: python projects are coming in, no SSH issues so far. Some people obviously don't read the docs, though :P
10:59
<dstufft>
antares_: lol wat are they doing
11:00
<antares_>
dstufft: just flipping the switch w/o adding .travis.yml
11:00
<antares_>
dstufft: so travis uses all the default settings: Ruby builder, Ruby 1.8.7, rake to run tests, etc
11:00
<josh-k>
reading takes too much time :)
11:01
<antares_>
josh-k: do you think we should rename php1,2 => ppp1,2?
11:01
<dstufft>
They can't contain their excitement long enough to read? ;)
11:01
<dstufft>
antares_: the queue still says PHP too, which i'm guessing PHP/PytohnPerl all go under it
11:02
<josh-k>
antares_: +1
11:02
<antares_>
josh-k: ah, right
11:02
<antares_>
josh-k: can we change queue title in the UI to say Perl, PHP, Python?
11:02
<antares_>
I will rename phpN => pppN hostnames later today
11:03
<josh-k>
antares_: yeah, i can change that
11:03
<jezdez>
josh-k: w0000t!
11:04
<antares_>
hey jezdez
11:04
<josh-k>
jezdez: roidrage was saying you might have some insight into the pip issue?
11:04
<jezdez>
josh-k: well, he just mentioned it'd be an issue of too many open connections to PyPI?
11:05
<jezdez>
can you be more specific what the error is?
11:05
<antares_>
jezdez: pip install opens so many network connections that it overwhelms our VM's NIC
11:05
<antares_>
so it starts dropping connections, including the ssh one worker uses
11:06
<antares_>
we will try to fix it on VM configuration end but it is interesting if we can limit # of requests pip is making
11:06
<antares_>
at a time
11:06
<jezdez>
antares_: threaded page scraping for the err win :(
11:06
<antares_>
page scraping??
11:06
<antares_>
jezdez: anyway, is there a way to configure this?
11:07
<jezdez>
pip doesn't have an option, but we're working on switching away from urlib to urllib3 which has connection pooling
11:07
<antares_>
ok, so it will take some time
11:07
<antares_>
josh-k: well, we will have to bring mitchell in and see if switching NIC adapter helps
11:08
<josh-k>
antares_: sweet
11:08
<jezdez>
antares_: we hope to have a release out in the next months indeed
11:08
<jezdez>
we'll sprint during pycon on it
11:09
<jezdez>
I'd argue that a builder image not being able to open connection is broken by default though
11:09
<jezdez>
despite the obvious bug in pip
11:09
<josh-k>
it is a fair thing to argue
11:10
<antares_>
jezdez: they are able to open them, but not hundreds at a time
11:10
<GitHub75>
[travis-ci] randym pushed 1 new commit to jasmine_patches: http://git.io/JKU7Fg
11:10
<GitHub75>
[travis-ci/jasmine_patches] events_spec cleanup - Randy Morgan
11:11
<antares_>
jezdez: we will see what we can do but this is the 1st time we seriously faced this issue
11:15
<antares_>
fortunately, so far this issue affects way less people than I was expecting
11:16
<roidrage>
sweet, stuff is running on rails 3.2.1
11:16
<GitHub64>
[travis-ci] mattmatt created rails-3.2 (+3 new commits): http://git.io/a77IRA
11:16
<GitHub64>
[travis-ci/rails-3.2] Bump libraries for Rails 3.2. - Mathias Meyer
11:16
<GitHub64>
[travis-ci/rails-3.2] Something for Heroku. - Mathias Meyer
11:16
<GitHub64>
[travis-ci/rails-3.2] Remove unused rake tasks. - Mathias Meyer
11:16
<travis-ci>
[travis-ci] travis-ci/travis-ci#1579 (jasmine_patches - 9bc2212 : Randy Morgan): The build passed.
11:16
<travis-ci>
[travis-ci] Change view : https://github.com/travis-ci/travis-ci/compare/48f3afc...9bc2212
11:16
<travis-ci>
[travis-ci] Build details : http://travis-ci.org/travis-ci/travis-ci/builds/751525
11:20
<josh-k>
kennethreitz: any good python sites to post to about this?
11:21
<kennethreitz>
josh-k: planet python
11:21
<josh-k>
kennethreitz: can you post there?
11:21
<kennethreitz>
josh-k: they only subscribe to rss though
11:21
<kennethreitz>
yeah I can
11:21
<josh-k>
kennethreitz: are there any big projects we can get on Travis and add the status badge to?
11:22
<travis-ci>
[travis-ci] travis-ci/travis-ci#1580 (rails-3.2 - 775e1ac : Mathias Meyer): The build passed.
11:22
<travis-ci>
[travis-ci] Change view : https://github.com/travis-ci/travis-ci/compare/681f374^...775e1ac
11:22
<travis-ci>
[travis-ci] Build details : http://travis-ci.org/travis-ci/travis-ci/builds/751581
11:22
<kennethreitz>
josh-k: oh they'll be coming
11:23
<antares_>
the taste of jenkins will bring them over
11:23
<antares_>
s/jenkins/setting up your own CI server
11:23
<kennethreitz>
antares_: that's dangerous talk :)
11:23
<antares_>
kennethreitz: I am not know for being polite or politically correct :)
11:24
<kennethreitz>
I don't think the two are mutually excuslive
11:24
<kennethreitz>
*exclusive
11:24
<kennethreitz>
Jenkins is *not* a test runner :)
11:24
<antares_>
kennethreitz: we actually recommend people who are asking about hosting a private travis install to use jenkins for now
11:25
<antares_>
kennethreitz: it is not and unfortunately, people keep asking for features that will make travis not a test runner
11:25
<jacobkm>
Hi folks - wondering what specs/costs are for a travis build machine. I'm thinking I'd like to donate a Python build slave.
11:26
<antares_>
jacobkm: hi
11:26
<kennethreitz>
^^ creator of django
11:26
<antares_>
jacobkm: let me pull up a link
11:26
<jacobkm>
antares_: hi!
11:26
<AutomatedTester>
I have posted a blog post about testing firefox addons on Travis CI if anyone is interest http://
www.theautomatedtester.co.uk/blog/2012/using-travis-ci-for-building-and-testing-firefox-addons.html
11:26
<AutomatedTester>
antares_: I made sure to mention that it needs to be done on the Node boxes :)
11:26
<antares_>
jacobkm: we use this for all new machines: http://
www.hetzner.de/en/hosting/produkte_rootserver/ex4
11:27
<jezdez>
jacobkm: *wave*
11:27
<antares_>
jacobkm: hetzner is where all but 1 worker machines are hosted. We transfer a ton of traffic over our internal network when we upgrade VM images so keeping them all colocated is a very good idea.
11:27
<jacobkm>
jezdez: hi, fancy running into you here :)
11:27
<antares_>
jacobkm: that said, we are trying to carefully add more machines. We probably don't really need one more right now. josh-k, what do you think?
11:28
<antares_>
AutomatedTester: sorry, I did not read your post, reading now
11:28
<AutomatedTester>
antares_: its cool, I can always update
11:28
<jacobkm>
antares_: I'm just thinking y'all are gonna get swamped by the incoming Python work and I'd like to help.
11:28
<josh-k>
antares_: jacobkm: can we hold off just for now but ping you about it at a later date? maybe if it really takes off for python we might need it, but the load is looking pretty good for now, wdyt?
11:29
<antares_>
AutomatedTester: links to travis-ci.org and our Getting Started guide would be very nice to have
11:29
<jacobkm>
josh-k: yeah totally -- please do ping me if you need help later.
11:29
<jacobkm>
jacob@jacobian.org
11:29
<antares_>
jacobkm: thank you, we will contact you when there's real need for another python machine
11:29
<AutomatedTester>
antares_: ok cool
11:30
<AutomatedTester>
antares_: the love link is good? :)
11:30
<josh-k>
jacobkm: thanks, i will make sure to ping you soon, if all goes to plan :)
11:30
<antares_>
AutomatedTester: yes, sure
11:30
<jacobkm>
It's fun watching all the Python projects trickle onto Travis this morning.
11:30
<iElectric>
guys.. where can I donate?
11:30
<iElectric>
seriously
11:31
<geekoncoffee>
https://love.travis-ci.org/
11:31
<antares_>
iElectric: love.travis-ci.org
11:32
<antares_>
AutomatedTester: other than that, your post looks good to me. Thank you!
11:32
<iElectric>
are open source repos going to be free?
11:32
<AutomatedTester>
antares_: no worries :0
11:32
<AutomatedTester>
:) even
11:32
<antares_>
iElectric: yes
11:32
<antares_>
iElectric: non-open source won't
11:32
<iElectric>
you just made the only java application I really HAD to use obsolete
11:33
<antares_>
svenfuchs, khaase: maybe we should update the love.travis-ci.org page to not mention future plans to add Python, Java and Scala :)
11:34
<josh-k>
iElectric: it is our pleasure :)
11:34
<antares_>
iElectric: several apps that make travis run are JVM-based but we feel your pain with that app ;)
11:34
<iElectric>
you are located in germany?
11:34
<josh-k>
iElectric: Amsterdam, Berlin and Moscow
11:35
<GitHub24>
[travis-ci] svenfuchs pushed 1 new commit to production: http://git.io/fT760g
11:35
<GitHub24>
[travis-ci/production] update the codeminer sponsor text - Sven Fuchs
11:35
<antares_>
the home base is in Germany, more or less
11:35
<geekoncoffee>
svenfuchs: you should also make the link to love more obvious... I clicked into "See our sponsors" and then manipulated the url to get there
11:35
<svenfuchs>
geekoncoffee: which link?
11:35
<geekoncoffee>
you should make it more obvious how to donate
11:35
<svenfuchs>
antares_: hmmm, well, yeah, maybe we should "tick" them off :)
11:36
<iElectric>
it's really pain to enter so much data to donate
11:36
<svenfuchs>
so we show what we've done
11:36
<iElectric>
but you deserve it
11:36
<antares_>
svenfuchs: yeah, even better
11:36
<svenfuchs>
geekoncoffee: we'll add a wikipedia-style "call for donations" at some point
11:36
<antares_>
arguably we already do it with notification emails
11:37
<josh-k>
svenfuchs: we should do a 'Jimmy Wales' style plea for donations
11:37
<josh-k>
svenfuchs: sexy pics of you asking for donations
11:37
<roidrage>
that's gonna be hot
11:38
<svenfuchs>
sexy pix of yehuda, aaron, etc
11:39
<iElectric>
tweet widget on "donation confirm" has two times inserted url
11:39
<iElectric>
just an fyi
11:39
<josh-k>
svenfuchs: sexy pics of Travis core
11:39
<josh-k>
svenfuchs: WE COULD DO A SWIMSUIT CALENDAR!
11:41
<svenfuchs>
iElectric: oops :) care enough to fix that?
11:42
<iElectric>
which repo?
11:43
<kennethreitz>
josh-k: I can post a full copy of the post
11:43
<kennethreitz>
josh-k: or an original one
11:44
<kennethreitz>
and i'm in a lazy mood
11:44
<travis-ci>
[travis-ci] travis-ci/travis-ci#1581 (production - 9017ef4 : Sven Fuchs): The build passed.
11:44
<travis-ci>
[travis-ci] Change view : https://github.com/travis-ci/travis-ci/compare/edd24d1...9017ef4
11:44
<travis-ci>
[travis-ci] Build details : http://travis-ci.org/travis-ci/travis-ci/builds/751716
11:45
<svenfuchs>
iElectric: https://github.com/travis-ci/travis-crowd
11:45
<svenfuchs>
iElectric: that would be awesome, we're kinda swamped right now :)
11:45
<iElectric>
kennethreitz: thanks for making best showcase for well designed api library
11:47
<josh-k>
kennethreitz: your pic :)
11:47
<kennethreitz>
iElectric: thanks for using it! (if you do :)
11:48
<kennethreitz>
josh-k: I'll copy but shorten, take out the perl stuff
11:48
<kennethreitz>
josh-k: be like "full details over here..."
11:49
<kennethreitz>
josh-k: heading out to lunch, so in an hour or two :)
11:49
<josh-k>
kennethreitz: <3<3<3
12:05
<iElectric>
how do I debug build with no output information?
12:05
<antares_>
iElectric: got a link?
12:06
<iElectric>
http://travis-ci.org/#!/iElectric/almir
12:06
<antares_>
the only iElectric/8* build I see has output
12:06
<AutomatedTester>
antares_: https://twitter.com/#!/AutomatedTester/status/174173652051439616 if you want to retweet with the travis account?
12:06
<antares_>
try refreshing the page (are you using opera by any chance?)
12:06
<antares_>
iElectric: «install: 'python bootstrap.py' returned false»
12:06
<iElectric>
antares_: where can I see that?
12:06
<antares_>
iElectric: http://travis-ci.org/#!/iElectric/almir/jobs/751868
12:07
<iElectric>
wasn't there before
12:07
<iElectric>
thanks :)
12:07
<antares_>
https://skitch.com/michaelklishin/8ffqh/travis-ci-distributed-build-platform-for-the-open-source-community
12:08
<antares_>
iElectric: we fixed most of UI issues that caused "blank output", maybe not all of them. Refreshing the page should help. Yes, like in 1998!
12:13
<antares_>
AutomatedTester: retweeted
12:13
<GitHub23>
[travis-ci] yumike opened issue #433: It seems that Python 2.5 support is broken. https://github.com/travis-ci/travis-ci/issues/433: http://git.io/g3eCqg
12:13
<antares_>
from wrong account
12:16
<svenfuchs>
antares_: i just noticed you're moving cookbooks issues to travis-cookbooks
12:16
<svenfuchs>
THANK YOU :)
12:16
<svenfuchs>
+1 on that
12:16
<antares_>
svenfuchs: well I don't personally care where they are but all CI env issues happen to be there
12:18
<svenfuchs>
yeah, it's becoming slightly overwhelming in travis-ci, so grouping them to a few repos makes sense to me :)
12:19
<antares_>
now it probably does
12:19
<antares_>
7 months ago we had an opposite discussion :)
12:20
<antares_>
well, you get what I am saying :)
12:20
<antares_>
AutomatedTester: retweeted as @travisci
12:24
<antares_>
AutomatedTester: I think you will like it: http://
www.flickr.com/photos/foca/6789331444
12:30
<ojii>
my (python) builds fail on python 2.5, but it works locally. looking at the output it looks like it mixes up 2.7 and 2.5. see http://travis-ci.org/#!/ojii/django-sekizai/jobs/752070
12:32
<AutomatedTester>
antares_: <3
12:33
<antares_>
ojii: what makes you believe it "mixed up 2.5 and 2.7"?
12:33
<ojii>
antares_, looking at the output of that there's a lot of "2.7 " in the output
12:33
<antares_>
ojii: it uses 2.7 in a virtualenv. 2.5 is not yet provided, we mentioned in the docs too soon.
12:34
<ojii>
ah, so no 2.5?
12:34
<antares_>
ojii: because 2.5 is not provided, your test suite must have picked up system python :(
12:34
<antares_>
which is 2.7
12:34
<antares_>
it will be added
12:34
<ojii>
btw can I run matrix builds with python? (in my case, different django versions)
12:34
<antares_>
ojii: watch https://github.com/travis-ci/travis-cookbooks/issues/29, it should be resolved in upcoming days
12:35
<antares_>
ojii: the same way you'd do it with Ruby or PHP or anything else, using ENV variables. See docs (I think PHP guide has some good examples)
12:35
<antares_>
but general guide should have them, too
12:36
<antares_>
ojii: see The Build Matrix at the bottom: http://about.travis-ci.org/docs/user/build-configuration/
12:36
<antares_>
env variables are permutated with runtime verisons, with a way to explicitly exclude some
12:37
<antares_>
ojii: our pleasure, good to see python projects on travis
12:37
<ojii>
can't wait to move everything over to travis
12:37
<ojii>
antares_, seeing there's no pricing page, this is free, right?
12:37
<antares_>
ojii: for OSS, it is
12:37
<antares_>
and always will be
12:39
<ojii>
so I finally have a proper reason to make the test runs for django cms saner ;-)
12:39
<ojii>
antares_, is there any support planned for publishing for example html files after the build (coverage reports) or processing xunit output?
12:39
<GitHub13>
[travis-ci.github.com] michaelklishin pushed 1 new commit to master: http://git.io/F5ywcw
12:39
<GitHub13>
[travis-ci.github.com/master] We documented Python 2.5 availability too soon. See https://github.com/travis-ci/travis-cookbooks/issues/29. - Michael S. Klishin
12:40
<antares_>
ojii: possibly. There is no complete consensus on that yet.
12:40
<antares_>
ojii: but quite likely, some day
12:41
<ojii>
especially coverage reports would be sweet
12:42
<ojii>
one more thing, can I disable the emails?
12:44
<deryl>
http://about.travis-ci.org/docs/user/build-configuration/
12:44
<deryl>
see Email Notifications
12:45
<ojii>
hrm imho should be in the web ui...
12:45
<antares_>
ojii: you can
12:45
<antares_>
ojii: pretty much everything is configured via .travis.yml
12:46
<antares_>
there will be parts that are configured on the Web but mostly for security-sensitive things
13:13
<antares_>
svenfuchs, josh-k: http://travis-ci.org/#!/facebook/tornado
13:13
<josh-k>
i have to tweet about that!
13:14
<antares_>
josh-k: and 2 out of 3 pythons actually pass, right away
13:15
<dreid>
Tornado has tests now?
13:18
<antares_>
josh-k: another python project I heard of: http://travis-ci.org/#!/webpy/webpy
13:18
<antares_>
I will add these to the docs
13:18
<sikachu>
Guys, I lost the link to the page for rvm rbx-18mode build script. Can I please have it again?
13:18
<antares_>
dreid: do you think simplejson can be considered a popular project?
13:18
<dreid>
antares_: Yes.
13:19
<antares_>
sikachu: https://github.com/travis-ci/travis-boxes/blob/master/config/worker.ruby.yml#L7
13:19
<antares_>
dreid: thanks, I will add it to examples as well
13:19
<sikachu>
antares_: thanks. I was looking at the wrong repo apparently (travis-worker)
13:19
<antares_>
sikachu: we have a few gazillions repos, it is not surprising :)
13:20
<josh-k>
dinner time
13:20
<sikachu>
josh-k: what time is it there O_o?
13:22
<GitHub131>
[travis-ci.github.com] michaelklishin pushed 1 new commit to master: http://git.io/E0OG_w
13:22
<GitHub131>
[travis-ci.github.com/master] A few more Python project examples: Tornado, simplejson, web.py - Michael S. Klishin
13:27
<GitHub41>
[travis-ci] michaelklishin pushed 1 new commit to master: http://git.io/cbYBZA
13:27
<GitHub41>
[travis-ci/master] Update the list of well-known projects - Michael Klishin
13:28
<GitHub180>
[travis-ci] michaelklishin pushed 1 new commit to master: http://git.io/YLkmYQ
13:28
<GitHub180>
[travis-ci/master] Link to travis-ci.org, stupid - Michael Klishin
13:30
<GitHub58>
[travis-ci] michaelklishin pushed 1 new commit to master: http://git.io/ngTLSA
13:30
<GitHub58>
[travis-ci/master] More popular Clojure projects that recently joined - Michael Klishin
13:32
<mmalecki>
antares_: hey, can I test a cookbook without building it?
13:32
<GitHub195>
[travis-ci.github.com] michaelklishin pushed 1 new commit to master: http://git.io/yk57PQ
13:32
<GitHub195>
[travis-ci.github.com/master] Only list what actualy Python community members think is worth listing - Michael S. Klishin
13:32
<mmalecki>
(I mean, without building a vm?)
13:33
<antares_>
mmalecki: well, you can run chef-solo locally and specify data bag and cookbook paths
13:33
<antares_>
but it is probably less convenient
13:33
<travis-ci>
[travis-ci] travis-ci/travis-ci#1582 (master - e191472 : Michael Klishin): The build passed.
13:33
<travis-ci>
[travis-ci] Change view : https://github.com/travis-ci/travis-ci/compare/48f3afc...e191472
13:33
<travis-ci>
[travis-ci] Build details : http://travis-ci.org/travis-ci/travis-ci/builds/752561
13:33
<mmalecki>
antares_: all right :)
13:33
<mmalecki>
antares_: working on ssh key cookbook
13:34
<antares_>
mmalecki: I can test something if you want. You don't have to build full VM images for production, just use sous-chef
13:34
<antares_>
I use sous-chef for all the cookbook development, not travis-boxes :)
13:34
<antares_>
github.com/michaelklishin/sous-chef
13:34
<mmalecki>
antares_: ok, I'll use it, thanks :)
13:36
<GitHub19>
[travis-lint] michaelklishin pushed 4 new commits to master: http://git.io/hOUDOw
13:36
<GitHub19>
[travis-lint/master] Quit with issues - Daan van Berkel
13:36
<GitHub19>
[travis-lint/master] Exit status spec - Daan van Berkel
13:36
<GitHub19>
[travis-lint/master] Captured stdout - Daan van Berkel
13:36
<antares_>
mmalecki: by the way, it was recently switched to vagrant 0.9. It works great with 0.8, too, just see Vagrantfile.sample comments.
13:36
<travis-ci>
[travis-ci] travis-ci/travis-ci#1584 (master - ded5b94 : Michael Klishin): The build passed.
13:36
<travis-ci>
[travis-ci] Change view : https://github.com/travis-ci/travis-ci/compare/d464b51...ded5b94
13:36
<travis-ci>
[travis-ci] Build details : http://travis-ci.org/travis-ci/travis-ci/builds/752570
13:36
kennethreitz joined
13:37
<mmalecki>
antares_: oh, awesome, I can go without 'bundle exec' :)
13:38
<antares_>
kennethreitz: can you please answer this tweet? I have no idea :) https://twitter.com/aaronsw/status/174199526209486848
13:38
<antares_>
mmalecki: if you bundle install with --bin-stubs, you can use ./bin/[executable], with any project managed by bundler
13:39
<travis-ci>
[travis-ci] travis-ci/travis-lint#15 (master - feb93a3 : Michael Klishin): The build passed.
13:39
<travis-ci>
[travis-ci] Change view : https://github.com/travis-ci/travis-lint/compare/8c82a1f...feb93a3
13:39
<travis-ci>
[travis-ci] Build details : http://travis-ci.org/travis-ci/travis-lint/builds/752600
13:40
<mmalecki>
antares_: should I develop ssh_keys cookbook in travis repo or in my fork? not sure if we want it back in the mainline
13:41
<antares_>
mmalecki: at this point we probably don't
13:41
<antares_>
mmalecki: see how you can combined multiple cookbook sets in the Vagrantfile.sample
13:41
<antares_>
this is what I use for my commercial projects, I sometimes develop some cookbooks side by side with OSS ones, including travis-cookbooks
13:42
<travis-ci>
[travis-ci] travis-ci/travis-ci#1583 (master - d464b51 : Michael Klishin): The build passed.
13:42
<travis-ci>
[travis-ci] Change view : https://github.com/travis-ci/travis-ci/compare/e191472...d464b51
13:42
<travis-ci>
[travis-ci] Build details : http://travis-ci.org/travis-ci/travis-ci/builds/752567
13:43
<mmalecki>
antares_: ok, so I guess I'll just create a repo for that :)
13:44
<dstufft>
antares_: that tweet confuses me
13:44
<dstufft>
http://packages.debian.org/squeeze/python-profiler <- the package he thinks he needs to install
13:44
<dstufft>
but profile and pstats are in the python stdlib..
13:45
<antares_>
dstufft: well, I replied this way: https://twitter.com/travisci/status/174203297144651776
13:45
<antares_>
dstufft: maybe you can take it from there. I really am not the right person to give any advice about python tools.
13:45
<dstufft>
apt-get'ing won't install the module into the virtualenv
13:46
<antares_>
dstufft: https://twitter.com/travisci/status/174203744488144896
13:47
<dstufft>
actually I think apt-get will work in this case
13:47
<dstufft>
because of what ubuntu did
13:48
<dstufft>
basically profile is a stdlib module, but it apparently has some licensing issues
13:48
<dstufft>
so debian/ubuntu split it out and put it in it's own repo
13:48
<dstufft>
s/repo/package/
13:48
<dstufft>
but their package is setup to drop it in the stdlib location
13:48
<dstufft>
so it should still work Ithink
13:48
<antares_>
and we on AngryOnRails the 2nd time: https://twitter.com/angryonrails/status/174201535776038913
13:49
<antares_>
dstufft: nice
13:49
<dstufft>
antares_: it might be useful to apt-get install python-profiler by default in the images, since it's normally part of stdlib
13:50
<dstufft>
next time you rebuild them of course
13:50
<antares_>
dstufft: hm, normally?
13:50
<antares_>
dstufft: ok, that's very easy to do
13:50
<dstufft>
antares_: ubuntu/debian apparently modify their python's stdlib to remove profile
13:51
<dstufft>
if you downloaded python from python.org and copiled it, it would have that package already
13:51
<dreid>
In all cases where your distribution is not run by people who really wanted to be Lawyers.
13:51
<dstufft>
debian cites licensing reasons and ubuntu just followed suit i'd assume
13:52
<dstufft>
http://serverfault.com/questions/123793/python-profiling-not-installed-in-ubuntu-how-do-i-get-it-in-a-virtualenv-and-wi
13:53
<antares_>
dstufft: I am really curious what's the reasoning for this
13:53
<antares_>
the typical "lets split everything into sub-packages" stuff debian is known for?
13:54
<antares_>
it is super annoying and confusing for Ruby packages, too
13:55
<dstufft>
http://packages.debian.org/squeeze/python-profiler
13:55
<dstufft>
"This package contains the modules 'profile' and 'pstats' from the standard library. They were move to a non-free package because of its license."
13:55
<antares_>
dstufft: thank you for helping us on this. I obviously would have no idea without you.
13:56
<antares_>
Python license is bsd-style, right?
13:56
<antares_>
anyway, not surprising for "freedom warriors"
13:58
<GitHub175>
[travis-cookbooks] michaelklishin pushed 1 new commit to master: http://git.io/FudHqA
13:58
<GitHub175>
[travis-cookbooks/master] Preinstall python-profiler package for PPP VMs - Michael S. Klishin
13:59
<dstufft>
antares_: it has it's own license (python), but it's similar to BSD yea
13:59
<dstufft>
I was under the impression that covered all the stdlib too, but I guess not
14:00
<antares_>
I see. Ruby has both Ruby License and BSD starting with 1.9.3
14:00
<dstufft>
antares_: and np for the help :)
14:03
<dstufft>
somewhere, someone is using code that isn't so free the mere sight of it makes money burst into flames -.-
14:17
<dstufft>
http://travis-ci.org/#!/fabric/fabric
14:20
<GitHub108>
[travis-ci.github.com] michaelklishin pushed 1 new commit to master: http://git.io/0aQbRA
14:20
<GitHub108>
[travis-ci.github.com/master] Update general Build Configuration guide to mention recently added languages - Michael S. Klishin
14:20
<antares_>
dstufft: worth providing as an example and listing in our "popular projects" list?
14:22
<GitHub194>
[travis-ci.github.com] michaelklishin pushed 1 new commit to master: http://git.io/DhSTvg
14:22
<GitHub194>
[travis-ci.github.com/master] Add fabric/fabric as an example - Michael S. Klishin
14:22
<dstufft>
antares_: https://crate.io/packages/Fabric/
14:22
<dstufft>
492,151 total downloads
14:23
<GitHub168>
[travis-ci] michaelklishin pushed 1 new commit to master: http://git.io/yMe46Q
14:23
<GitHub168>
[travis-ci/master] Update README.textile - Michael Klishin
14:23
<antares_>
dstufft: done!
14:25
<dstufft>
antares_: 2.5 seems to be broken
14:26
<antares_>
dstufft: there is already a ticket for that and we even tweeted our apologies :)
14:26
<dstufft>
antares_: http://d.stufft.io/1C1k0X0Z460w133N3h3d bitprohpet == fabric guys
14:26
<antares_>
dstufft: and I removed it fro mthe docs for now
14:26
<dstufft>
antares_: ah oops
14:27
<antares_>
dstufft: his understanding is correct, we did not deploy new PPP VM images that have 2.5
14:27
<antares_>
I am checking that there are no issues with 2.5 this very second
14:27
<antares_>
dstufft: how do I check that python-profiler works?
14:28
<dstufft>
python -c "import profile; import pstats"
14:28
<dstufft>
if you get no output it works
14:28
<travis-ci>
[travis-ci] travis-ci/travis-ci#1585 (master - 9892d46 : Michael Klishin): The build passed.
14:28
<travis-ci>
[travis-ci] Change view : https://github.com/travis-ci/travis-ci/compare/ded5b94...9892d46
14:28
<travis-ci>
[travis-ci] Build details : http://travis-ci.org/travis-ci/travis-ci/builds/752904
14:28
<dstufft>
if you get an ImportError it doesn't
14:28
<antares_>
do I import profiler?
14:29
<* antares_>
launches python shell every 3 years or so :(
14:29
<dstufft>
nope, just what I typed
14:29
<dstufft>
jsut copy/paste that to shell
14:30
<thepumpkin>
antares_: +1
14:30
<dstufft>
antares_: http://d.stufft.io/0N161x380y2K0t3l0L34 again the fabric guy ;)
14:30
<antares_>
thepumpkin: what are you +1ing? :)
14:30
<thepumpkin>
antares_: launching python shell every 3 years haha
14:31
<antares_>
dstufft: please point this person to http://about.travis-ci.org/docs/user/languages/python/ and http://about.travis-ci.org/docs/user/ci-environment/
14:31
<antares_>
dstufft: what channel is that?
14:31
<thepumpkin>
Last time I did python was in 2009 for AppEngine.
14:31
<dstufft>
antares_: first screenshot was #django-social, second was a PM
14:34
<antares_>
dstufft: I just realized that python-profiler and python2.5 clash :/
14:34
<antares_>
dstufft: https://gist.github.com/444b95cbf087515e9899
14:35
<dstufft>
ubuntuuuuu
14:35
<dstufft>
and/or debiannnnnn
14:35
<dstufft>
so apparently the python-profiler still has the files for python2.5
14:35
<GitHub23>
[travis-boxes] joshk pushed 1 new commit to master: http://git.io/t2fy4Q
14:35
<GitHub23>
[travis-boxes/master] change the NIC type used by the VM as the default one starts dropping connections when it gets swampped (see https://github.com/mitchellh/vagrant/issues/516) - Josh Kalderimis
14:36
<dstufft>
even though ubuntu doesn't support it anymore
14:36
<dstufft>
and deadsnakes didn't remove python2.5's profile and pstats
14:36
<josh-k>
antares_: Mitchell confirmed that the NIC fix has had good results for users
14:36
<antares_>
josh-k: phew
14:36
<josh-k>
i have added it to travis boxes and the vagrant worker commands
14:36
<antares_>
dstufft: hm, so what do we do?
14:36
<dstufft>
antares_: trying to think :3
14:37
<antares_>
josh-k: cool. I suggest that we deploy it for ppp VMs as soon as you think it's ready and for ruby VMs tomorrow. It will take full VM upgrade.
14:37
<dstufft>
fucking debian screwing shit up
14:37
<antares_>
josh-k: also, I am working on some python CI environment issues, should be done soon
14:37
<josh-k>
antares_: we need to test the worker updates on staging as I had to fork and update net-ssh-shell
14:37
<dstufft>
antares_: for right now i'd leave python-profiler off
14:38
<dstufft>
something will have to be figured out with that
14:38
<dstufft>
maybe just manually dropping the profile.py and pstats.py into the stdlib
14:38
<dstufft>
or something
14:38
<dstufft>
might be able to just modify the deb
14:38
<dstufft>
and remove python2.5
14:38
<antares_>
that sounds like something that will require a lot of testing and careful consideration :)
14:39
<dstufft>
yea basically
14:39
<dstufft>
it's a bad situation
14:39
<dstufft>
kennethreitz: might have a better idea, but i'll think on it
14:40
<antares_>
dstufft: I am afraid we will have to do without python-profiler for now :(
14:40
<antares_>
let me google and search apt-cache
14:41
<GitHub74>
[travis-cookbooks] michaelklishin pushed 1 new commit to master: http://git.io/8M3LMg
14:41
<GitHub74>
[travis-cookbooks/master] Damn debian/ubuntu still distributes python2.5 files in the python-profiler package - Michael S. Klishin
14:42
<kennethreitz>
dstufft: why is that neccesary?
14:42
<antares_>
dstufft: status: new, importance: undecided: https://bugs.launchpad.net/ubuntu/+source/python-profiler/+bug/809299
14:42
<antares_>
dstufft: they are undecided about stdlib bits not being available or even installable for 2.6 :)
14:43
<dstufft>
kennethreitz: debian/ubuntu decided to remove parts of stdlib
14:43
<dstufft>
there is a package to add it back in
14:43
<dstufft>
but that package clashes with deadsnakes 2.5
14:43
<dstufft>
because that package still has the files for 2.5
14:43
<dstufft>
but deadsnakes didnt remove those files
14:43
<dstufft>
so apt errors out
14:44
<kennethreitz>
dstufft: um wtf
14:44
<kennethreitz>
why?
14:44
<dstufft>
oh why did they remove parts of stdlib?
14:44
<dstufft>
because FREEDOM
14:44
<antares_>
kennethreitz: freedom first, common sense second (if at all)
14:44
<dstufft>
http://packages.debian.org/squeeze/python-profiler
14:45
<kennethreitz>
distros have killed python
14:45
<dstufft>
the only options I can think of are a) manually remove those files from 2.5 and let python-profiler install them again (UGH), b) manually add those files to 2.6+ and forget python-profiler (UGH) and c) go all custom PPA on python
14:46
<dstufft>
forgot to say (UGH) after c
14:46
<dstufft>
but it's there
14:46
<dstufft>
in my brain
14:46
<kennethreitz>
dstufft: that looks like an open license to me
14:46
<antares_>
dstufft: we can remove files (we do it to fix the issue we had with 2.6 before) but it gets out of hand very quickly
14:47
<antares_>
Chef has very nice system of events, too, but having 3-4 actions triggered by events makes a very straightforward cookbook a hell :(
14:47
<dstufft>
kennethreitz: I think what they did is stupid
14:47
<dstufft>
and wrong
14:47
<dstufft>
but arguing with debian is not on my todo list today
14:48
<dstufft>
catalog-sig is about all I can handle on my smash face against desk scale
14:54
<kennethreitz>
oh i can't come close to that
14:55
<josh-k>
antares_: and NOW Rails decides to push lots
14:56
<antares_>
josh-k: :) well, if they complain we will figure something out
14:56
<josh-k>
no worries :)
14:56
<josh-k>
80 new repos today!
14:56
<antares_>
ruby4 will stay ruby4 but I guess I won't be shy of adding one more machine like I originally suggested with jvm1 :)
14:57
<antares_>
josh-k: I am putting up some more JVM-based ones today if all goes well :)
14:57
<antares_>
josh-k: also, we updated the README: github.com/travis-ci/travis-ci
14:57
<antares_>
josh-k: and I think we should consider updating love.travis-ci.org, too
14:57
<josh-k>
with what?
14:58
<josh-k>
what did you add to the readme?
14:59
<antares_>
josh-k: Python, Scala, Clojure, Java projects
15:00
<antares_>
josh-k: more Clojure, completely new lists for the others
15:00
<antares_>
josh-k: just bombarding visitors with titles and buzzwords, nothing special :)
15:04
<dstufft>
option D) package profile and pstats into it's own module that can be pip installed
15:04
<dstufft>
probably the best option
15:04
<dstufft>
it should work jsut fine
15:04
<antares_>
dstufft: and something that can be maintained by python developers, not us :)
15:24
<antares_>
josh-k, dstufft : are you on quora? Maybe you can leave a comment about our announcement here? http://
www.quora.com/Are-there-any-Python-Django-friendly-hosted-Continuous-Integration-services
15:25
<dstufft>
im not on quora
15:25
<josh-k>
i am on it but i have never used it
15:27
<antares_>
I don't have an account and I think quora is invite-only
15:28
<iElectric>
right now postfix wants to configure
15:29
<antares_>
iElectric: sorry?
15:29
<iElectric>
http://travis-ci.org/#!/iElectric/almir/jobs/753322
15:29
<antares_>
iElectric: you want to send email from travis?
15:29
<iElectric>
no, package depends on it
15:29
<iElectric>
any idea which parameter to send to apt-get?
15:30
<antares_>
iElectric: we configure apt-get to assume -y by default
15:30
<antares_>
what postfix needs is called preseeding
15:30
<iElectric>
maybe it needs --really-y
15:30
<antares_>
I am not sure how to pass preseeding values from the command line, they are usually taken from a file
15:31
<iElectric>
maybe I can tell aptitude not to install that dependency
15:32
<dstufft>
antares_: a really long echo statement? ;P
15:32
<dstufft>
echo " lol " > preseedfile
15:33
<antares_>
iElectric: try --nodeps and --ignore-depends (for apt-get install)
15:33
<iElectric>
or install sendmail
15:33
<iElectric>
that may satisfy mta slot
15:33
<antares_>
iElectric: sorry, --ignore-depends is for dpkg
15:33
<antares_>
iElectric: just curious what does your project do?
15:34
<antares_>
iElectric: for obvious reasons we block outgoing SMTP traffic in the host OS
15:34
<iElectric>
I don't need the emails
15:34
<iElectric>
bacula pulls postfix as dependency
15:34
<antares_>
I understand, but why do you install something that depends on postfix :/
15:34
<iElectric>
because I am writing web interface for it
15:35
<dstufft>
antares_: you mean I can't use travis to send out my great notices about free "enhancements"
15:35
<antares_>
iElectric: ah, ok
15:35
<antares_>
dstufft: you can try but we will do our best to make sure they do not get delivered :(
15:35
<antares_>
dstufft: one gentleman tried. We had to contact his boss :(
15:36
<iElectric>
then the boss tried?
15:36
<antares_>
the project said something like "hello cloud hipsters". So now we know who we are.
15:36
<antares_>
iElectric: hahaha
15:36
<iElectric>
you do now know? so he succeeded ;)
15:37
<antares_>
iElectric: it was in the github project description
15:37
<antares_>
iElectric: emails only had "lol" in them. It was called lol-cloud or something.
15:38
<antares_>
iElectric: and yes, it was a wake up call for us. It was months ago, fortunately.
15:38
<iElectric>
this thing is pretty solid
15:38
<antares_>
I think we just added php support back then (October 2011), so naturally some creative php developers had to build a lol-cloud.
15:38
<iElectric>
first we got readthedocs.org
15:38
<iElectric>
now travis
15:38
<iElectric>
someone does agile payments and we might have agile IT ;)
15:39
<antares_>
iElectric: I heard stripe is good, braintree is also nice
15:39
<AutomatedTester>
iElectric: stripe?
15:40
<iElectric>
but they don't integrate into github
15:40
<iElectric>
we need pledges for tickets
15:44
<antares_>
iElectric: ahh
15:45
<dstufft>
"lodge a ticket? pay me and i'll fix it"
15:45
<iElectric>
It would so work.
15:45
<iElectric>
still a student, if no one does it in next 2 years
15:45
<iElectric>
I'm so for it.
15:46
<mmalecki>
http://gun.io/ has something like this
15:46
<mmalecki>
but agreed, it'd be *so* awesome
15:47
<iElectric>
but this is still job based
15:47
<iElectric>
we are talking micro, tickets
15:47
<mmalecki>
http://gun.io/open/
15:47
<mmalecki>
some are tickets
15:48
<antares_>
so people chip in to pay for issue resolved?
15:48
<antares_>
or what tickets are we talking about?
15:48
<iElectric>
exactly
15:48
<antares_>
very nice
15:48
<iElectric>
mmalecki: nice, this is very close to it
15:50
<iElectric>
US only, as always
15:50
<antares_>
gun.io says I committed -1 day ago
15:51
<antares_>
good to know
15:51
<mmalecki>
iElectric: yeah, there are some crazy law restrictions
15:51
<iElectric>
btw, UX of travis is really wierd, can I fire up a ticket somewhere?
15:51
<iElectric>
or just shout here
15:51
<mmalecki>
damn antares_. you're going over the speed of light!
15:51
<iElectric>
mmalecki: I can imagine
15:51
<antares_>
the UI hasn't changed much since the very beginning and there are some improvements we need but "weird" is not specific enough
15:52
<antares_>
iElectric: svenfuchs is our UI mastermind
15:52
<iElectric>
svenfuchs: around? :)
15:53
<mmalecki>
I silently hope to see ember.js gone from the UI one day ;)
15:54
<josh-k>
mmalecki: why?
15:54
<mmalecki>
josh-k: I hate all kinds of monolithic frameworks
15:55
<mmalecki>
this is just my personal aversion, so nothing important
15:55
<josh-k>
mmalecki: i think you are over simplifying it :)
15:55
<mmalecki>
josh-k: most likely I am!
15:57
<mmalecki>
(also I'm biased)
15:57
<mmalecki>
but you know, I'm just that weird guy over there hatin' on everything
15:58
<antares_>
mmalecki: we used some other stuff before Ember. I have never worked on the UI but my impression is that Ember does provide enough to bear with some of its issues.
16:02
<mmalecki>
antares_: I'm aware of all the functionalities Ember provides and I know how they are used in Travis. However, in my opinion, all this stuff could be replaced with something smaller and modular. I feel that Ember breaks few JS patterns and it's poor documentation increases entry level.
16:03
<antares_>
mmalecki: documentation is definitely an issue. That said, Ember is not abandoned.
16:04
<antares_>
mmalecki: and sometimes writing docs takes as much time as rewriting the project from scratch. http://rubyamqp.info did :)
16:04
<antares_>
so I personally trust svenfuchs' choice, even though I am a huge fan of MooTools ;)
16:04
<antares_>
speaking of modular ;)
16:04
<antares_>
mmalecki: do you like mootools approach better?
16:06
<mmalecki>
antares_: I know writing documentation is hard. We're trying to deal with it in flatiron now, but each flatiron module is *really* tiny, so going through code takes no more than few minutes. And yes, mootools > Ember :)
16:06
<antares_>
mmalecki: but good docs are not just inline comments/reference
16:06
<antares_>
any serious project needs guides
16:07
<antares_>
and those need introduction to be written, guides to link to each other, make sense overall, CSS work, etc
16:07
<antares_>
it is a lot like writing a book
16:07
<antares_>
but I see what you're saying
16:07
<antares_>
I personally hope we will have a chance to seriously contribute to projects we use
16:07
<mmalecki>
antares_: oh, it's in the works. and it's going to be awesome :)
16:09
<antares_>
mmalecki: if I had to evaluate node.js today for web apps, would you recommend flatiron?
16:09
<antares_>
mmalecki: if so, I am looking forward to it :)
16:09
<mmalecki>
antares_: yes, I definitely would. it's been in use in production for few months already.
16:10
<mmalecki>
antares_: it's just missing some pieces of documentation about integrating it correctly, but the point of flatiron is that there's no "only correct way"
16:10
kennethreitz joined
16:10
<mmalecki>
antares_: also, nodejitsu support :)
16:14
<warreng>
is it possible to setup travis-ci on my own servers and configure it to use private github repos?
16:14
<antares_>
warreng: it is possible but is not easy and things change every week
16:15
<antares_>
warreng: so we recommend sticking to jenkins until travis-ci.com opens its doors
16:15
<josh-k>
warreng: Travis has so many moving parts due to its distributed nature
16:15
<josh-k>
warreng: also, it is hard coded for public projects only at the moment
16:16
<josh-k>
antares_: 90 projects today :)
16:16
<warreng>
so jenkins is the recommended CI otherwise?
16:16
<josh-k>
warreng: well, for now
16:17
<mmalecki>
antares_: as a nice example of flatiron flexibility, here's a presentation tool using broadway: https://github.com/mmalecki/presents
16:17
<antares_>
josh-k: I think what names we added today is even more important. Plenty of Python projects I heard of, some interesting non-Python projects. I submitted PRs for the most popular Clojure web framework and data mapping/ODM kind of library.
16:18
<mmalecki>
and a slideshow made using it http://flatiron-meetjs-summit-2012.jit.su/
16:18
<antares_>
the rising (Python) wave lifts all boats
16:19
<antares_>
mmalecki: what does "isomorphic" mean for a library?
16:20
<mmalecki>
antares_: ability to use it in a browser and on a server (node.js)
16:20
<antares_>
my math degree is screaming "wrong! wrong!" :)
16:21
<antares_>
mmalecki: looking forward to catamorphic JavaScript frameworks!
16:21
<mmalecki>
I know, someone pointed it our already :)
16:22
<mmalecki>
but well, it's just a word which happens to have a different meaning in maths and in programming
16:22
<antares_>
several different meanings in math alone :)
16:27
<antares_>
loicfrering: https://twitter.com/vtourraine/status/174243623238516736 :)
16:27
<antares_>
loicfrering: is this really true? :)
16:30
<antares_>
josh-k: I fixed some minor issues in the python recipe, when do you think we should build/deploy new VMs?
16:30
<josh-k>
sure, just not with vagrant 0.9 yet :)
16:30
<antares_>
josh-k: switching NIC will take pretty much a full VM upgrade
16:31
<antares_>
ok, so I will deploy them and then we will see about NICs
16:42
<soupmatt>
hello, I'm having an issue getting my project to pass under jruby-19mode on travis
16:42
<soupmatt>
here's the link to the failing build http://travis-ci.org/#!/rubycas/rubycas-client/jobs/753924
16:43
<soupmatt>
it works fine under jruby-18mode, and I can run it with --19 on my laptop just fine as well
16:43
<antares_>
soupmatt: does it use DB connection?
16:44
<antares_>
soupmatt: we have seen cases when people had to use "127.0.0.1" instead of "localhost" and "root" instead of blank username with the mysql JDBC driver
16:44
<antares_>
nothing else I can name immediately. Maybe ask on the JRuby mailing list?
16:45
<antares_>
also, if you use OS X locally, then test execution order is likely to be different on Linux
16:45
<antares_>
because Dir.glob and friends return files in different order
16:45
<antares_>
this sometimes makes a difference for test suites
16:47
<soupmatt>
I haven't tried mysql, I've just been using h2 and sqlite3 at this point, so there isn't any network involved. I'll ask the jruby guys as you suggest.
16:47
<soupmatt>
thx for the help :)
16:47
<antares_>
soupmatt: interesting.
16:48
<antares_>
soupmatt: we can also give you a URL to our Ruby VM image if you want
16:48
<antares_>
soupmatt: you'll have to use vagrantup.com and it is 3.1 GB in size but if you want :)
16:49
<soupmatt>
antares_: I'm not that desperate yet
16:49
<soupmatt>
but thanks for the offer
17:02
<josh-k>
oh dukeleto
17:02
<josh-k>
why you flood us like this :)
17:04
<josh-k>
antares_: we got 99 projects today, and my code ain't one
17:04
<antares_>
josh-k: I thought 98 of them were yours!
17:05
<antares_>
hopefully 2 of my PRs will be accepted and we will get over 100
17:05
<antares_>
I don't remember us going over 100 projects in 1 day
17:05
<antares_>
even after Rails move we added something like 60
17:05
<antares_>
I am mostly happy that NIC issues do not affect most of people
17:05
<antares_>
I was seriously worried about that
17:07
<josh-k>
antares_: can you add me to root for staging please
17:07
<josh-k>
i am trying to test travis-boxes but i get gem install issues
17:08
<antares_>
josh-k: try now
17:08
<antares_>
I think we have only one machine where rvm is installed system-wide
17:08
<antares_>
by accident
17:08
<antares_>
jvm1, I think
17:09
<antares_>
what gem fails to install?
17:09
<josh-k>
libxml-ruby
17:10
<antares_>
I see. Yes, we probably don't have libxml on staging.
17:10
<iElectric>
is there a way to turn off emails to organizations?
17:12
<antares_>
iElectric: yes, http://about.travis-ci.org/docs/user/build-configuration/ describes how to disable email or all notifications
17:12
<josh-k>
libxml2 is on staging, i think it is an rvm issue
17:13
<antares_>
josh-k: do you want me to take a look?
17:13
<iElectric>
antares_: but it does not say how to disable emails for organizations
17:13
<josh-k>
that would be great, it works fine locally
17:13
<iElectric>
I still want git commiter to get the emails
17:13
<antares_>
iElectric: specify recipients explicitly. Right now this is the only option.
17:14
<antares_>
iElectric: in part historically, in part because github API has quirks about who is org repo "owner"
17:14
<iElectric>
then subquestion would be, travis caches organization email right?
17:14
<iElectric>
because I removed it from github, but it will still send it
17:14
<antares_>
iElectric: hm, for what?
17:14
<antares_>
we don't cache emails but if you don't list recipients explicitly, we obtain emails from github
17:15
<iElectric>
I removed email from organization profile on github
17:15
<antares_>
give me a second
17:15
<iElectric>
but it is still sending it
17:15
<antares_>
iElectric: regardless of why, right now for organizations your best option is to list recipients explicitly
17:16
<antares_>
overhaul of some email notification defaults is definitely on the shortlist, we hear questions and complaints about it often enough
17:16
<antares_>
iElectric: yes, owner_email is stored in the repos tables
17:17
<josh-k>
iElectric: antares_ is 120% right, we need to overhaul some of it, but it won't just next week :/
17:17
<antares_>
but I cannot immediately tell why and if we still need to do it
17:18
<iElectric>
I totally understand
17:19
<iElectric>
wish I knew ruby :-)
17:20
<josh-k>
always a good time to learn :)
17:20
<mmalecki>
iElectric: you totally should learn Ruby! it's a nice language
17:21
<mmalecki>
also, just for the sake of learning :)
17:22
<iElectric>
that's the problem, I'm too happy with python
17:22
<iElectric>
kills my motivation to learn anything else
17:22
<josh-k>
i would like to learn some python someday
17:22
<josh-k>
just wish i had more time
17:23
<iElectric>
something along those words, yes
17:23
<mmalecki>
oh come on, learning a new language takes one day at most!
17:23
<mmalecki>
you just need the right method
17:24
<mmalecki>
which is: start a fucking project
17:24
<josh-k>
mmalecki: tut tut tut
17:24
<josh-k>
mmalecki: i am old now
17:24
<josh-k>
takes me ages
17:24
<mmalecki>
josh-k: oh... you should try being young then!
17:24
<josh-k>
and i am a little busy with Travis :)
17:24
<antares_>
mmalecki: I'd say Python and Ruby are similar enough to not teach you much, Ruby has more Lisp influence and DSLs to learn from but they have more or less the same applicability and limitations :)
17:25
<dgeb>
I'm looking for advice on configuring a .travis.yml for use with phantomjs. Is it possible to use the file protocol to access a simple qunit test?
17:25
<antares_>
dgeb: we usually see people starting small webservers to serve pages, usually in Ruby or node.js
17:26
<iElectric>
python -m SimpleHTTPServer
17:26
<mmalecki>
antares_: python seems more liberal to me in terms of oop, but yeah, agreed
17:26
<antares_>
dgeb: it may be possible (we are not phantom.js experts, unfortunately) but I can't think of any examples
17:26
<iElectric>
it will fire of a server in current folder
17:26
<mmalecki>
`http-server` :)
17:26
<dgeb>
antares_: ok - thanks.
17:27
<dgeb>
perhaps the files aren't in predictable locations, so it's safer to start a server? i was just trying to keep this as lean as possible
17:27
<antares_>
mmalecki: sure but unless I have a reason to use it, I won't learn myself Python. I am sure it is worth learning but things I work on (including travis) would benefit more from my C and Erlang skills going up :)
17:28
<antares_>
dgeb: I'd says so, although if you know github project name, git clone location is predictable
17:28
<iElectric>
I would go with haskell, just to fuckup the brain a little more
17:29
<antares_>
iElectric: I have been working in Clojure for 7+ months now :)
17:29
<iElectric>
any fun?
17:29
<josh-k>
antares_: any idea on the libxml-ruby gem issue?
17:29
<dgeb>
antares_: do you know the full path to the git clone location?
17:30
<antares_>
iElectric: yes, I absolutely love how much it lets me focus on the problem and how much easier certain things become with immutable data structures
17:30
<antares_>
josh-k: what do I need to do to reproduce?
17:30
<antares_>
josh-k: just install nokogiri or something?
17:30
<josh-k>
gem install libxml-ruby
17:31
<antares_>
dgeb: if your project slug is abc/xyz, git clone will be located at ~/builds/abc/xyz
17:32
<dgeb>
antares_: ah, yes, i see that in the build. thanks much!
17:33
<antares_>
josh-k: ok, I see it failing. Investigating.
17:35
<antares_>
josh-k: it fails (although gets quite far) on jruby and installs just fine on 1.9.2 now
17:35
<antares_>
josh-k: libxml2-dev and libxml2-utils were not installed
17:35
<antares_>
can it be that you are trying on jruby?
17:36
<josh-k>
i was trying on 1.9.3
17:37
<josh-k>
and i tried on jruby too
17:37
<antares_>
josh-k: I just installed it on 1.9.3, try it
17:37
<josh-k>
yep, installed fine now on 1.9.3
17:37
<josh-k>
damn, fails big time on jruby
17:38
<josh-k>
ok, only use travis boxes with 1.9.3
17:38
<antares_>
josh-k: well, it is a C extension. Relying on them on JRuby should be last resort. What depends on libxml-ruby, do you know?
17:38
<josh-k>
VeeWee -> CFPropertyList -> libxml-ruby
17:38
<antares_>
I always use 1.9.3 with travis-boxes locally and 1.9.2 (because you recommended 1.9.2 in 1.9.3-p0 days) on all of our production machines
17:38
<josh-k>
it should use nokogiri and it would be fine
17:39
<antares_>
ah, damn. veewee has OS X property lists support?
17:39
<josh-k>
i heard there will be another 1.9.3 release soon
17:39
<antares_>
well, it should be pretty trivial to switch it to nokogiri
17:39
<antares_>
nokogiri has JRuby version
17:40
<josh-k>
which is nice
17:40
<josh-k>
Mitchell is working on a new ruby ssh client for vagrant, but it will have bindings to a c ext, so i doubt it will work with jruby well
17:41
<antares_>
josh-k: I know. For JRuby there is sshj we can use
17:41
<antares_>
in fact, it is significantly less work to wrap sshj than libssh
17:41
<antares_>
maybe we can discuss doing a JRuby version, we will see
17:41
<josh-k>
antares_: staging has the updating t-b with vagrant 0.9, wanna try it out?
17:42
<antares_>
I want to see a replacement to net/ssh this year, travis will need it badly, especially travis pro
17:42
<antares_>
josh-k: I can build a new ppp image with it
17:42
<josh-k>
go ahead :)
17:42
<antares_>
or a new jvm image, there is something to deploy as well
17:42
<josh-k>
and +1 to working with Mitchell on a java/jruby ssh lib
17:45
<antares_>
josh-k: how can I contact mitchell? I usually don't find him in #vagrant with our 12 hours difference
17:46
<antares_>
josh-k: if you can, maybe he can add me in skype? I have been investigating sshj for a while now
17:46
<antares_>
my only problem is how to write tests for this thing :(
17:46
<antares_>
but sshj has tests
17:46
<antares_>
hm, why isn't it on travis, by the way :/
17:46
<josh-k>
i have him on IM, i will connect you both up via email
17:47
<josh-k>
it isn't os yet
17:49
<antares_>
josh-k: ok, thanks
17:58
<antares_>
josh-k: I am build a new ppp image on staging
17:58
<josh-k>
once you are done then i will test the changes i made to the worker on staging
17:59
<josh-k>
have we updated all workers to jruby 1.6.7?
17:59
<antares_>
josh-k: the same thing, Base MAC address for eth0/NAT must be set. Contact box maintainer for more information
17:59
<antares_>
josh-k: no, not yet. We will this week.
18:00
<josh-k>
i have a fix for that
18:00
<josh-k>
let me make sure i added it
18:00
<antares_>
I am in box-building/travis-boxes, git repo is up to date
18:00
<antares_>
I also pulled box-building/travis-cookbooks
18:01
<josh-k>
antares_: ah, i missed something, one sec
18:02
<GitHub63>
[travis-boxes] joshk pushed 1 new commit to master: http://git.io/HlX_Tg
18:02
<GitHub63>
[travis-boxes/master] need to add a vagrant.reload! after adding a box - Josh Kalderimis
18:02
<josh-k>
antares_: done, you can pull
18:09
<antares_>
josh-k: with vagrant 0.9 chef output is all green :/
18:09
<antares_>
I feel like I am finally in the matrix
18:11
<josh-k>
antares_: hahaha, i can make it grey
18:13
<antares_>
josh-k: it is configurable?
18:14
<josh-k>
yep, in vagrant.rb
18:14
<antares_>
all previous vagrant versions used white. I don't care about the color as long as I can read Chef errors in red but I am not very used to all green all the time.
18:16
<josh-k>
hahaha, ok
18:16
<josh-k>
this is just vagrant color output
18:20
<antares_>
josh-k: loicfrering coinced a new term, TDD = Travis Driven Development
18:21
<josh-k>
antares_: i used that in New York for SpreeConf
18:21
<antares_>
old news then
18:21
<robgleeson>
antares_: I think that's already taken :P
18:21
<josh-k>
I renamed my presentation Travis Driven Dev :)
18:22
<AutomatedTester>
hey guys, whats the best way to see why something can't connect to the mysql during a run?
18:23
<AutomatedTester>
btw antares_ my blog post from earlier was picked up by the team that builds the Firefox Addon SDK and he was talking about how this could be a "best practise"
18:24
<AutomatedTester>
he will let me know more tomorrow after the team meeting
18:24
<antares_>
AutomatedTester: there is no good way except for one of us to do it
18:24
<antares_>
AutomatedTester: is this a node.js project or?
18:25
<AutomatedTester>
antares_: yea
18:26
<antares_>
AutomatedTester: what host and username do you use to connect?
18:27
<antares_>
AutomatedTester: try 127.0.0.1 instead of localhost and root instead of empty/default username
18:33
<AutomatedTester>
antares_: ok will try that
18:54
<AutomatedTester>
antares_: being explicit on a few things sorted it
18:55
<antares_>
AutomatedTester: can you explain what exactly you had to change in case someone else hits this issue?
18:55
<AutomatedTester>
lemme find the patch
18:55
<AutomatedTester>
antares_: https://github.com/lloyd/browserid/commit/1d6e86a3d3d26a6f7a6e928f8710323dd932064c
18:55
<antares_>
AutomatedTester: yeah, it must be the username
18:56
<antares_>
Java JDBC driver and one of Ruby clients also have this tricky aspect
18:56
<AutomatedTester>
and the creating a DB
18:56
<antares_>
AutomatedTester: haha, a mongodb/couchdb habit? ;)
18:56
<AutomatedTester>
antares_: I have a nasty mongodb habit I cant kick
18:57
<AutomatedTester>
antares_: the tests should create the DB so its a bug in our tests
18:57
<AutomatedTester>
but we have a green build for us now to break things
18:57
<antares_>
AutomatedTester: you need a 3 day long SQL rehab course :)
18:57
<AutomatedTester>
only if the course is with Access
18:57
<* AutomatedTester>
wants proper pain
18:57
<AutomatedTester>
:P
19:32
kennethreitz joined
19:41
<binbasti>
galfert: liquor cabinet is jetzt auf travis und grün
19:41
<binbasti>
using the wrong irc windows all the time
20:02
geekoncoffee joined
20:14
jaysonsantos joined
20:16
<jaysonsantos>
Hi guys, how you doing? I'm enjoying travis for python and I already have 2 projects using it, however one of of those is a memcached cliente which tests with server listening on unix socket will not pass. Is there anyway to test it with unix socket?
20:24
<antares_zzzz>
jaysonsantos: can memcached listed on both TCP and UNIX socket at the same time?
20:25
<antares_zzzz>
jaysonsantos: if it can, we can tweak our CI environment. If not, you will have to do something in your test suite. You have passwordless sudo and VMs are snapshotted so feel free to do anything you want but it may or may not be easy.
20:28
<jaysonsantos>
antares_zzzz it only listen in one port at time, if it is possible, i can run it 3 times as local user to listen in default port, another port and unix socket
20:28
<antares_zzzz>
jaysonsantos: you can stop it, copy your own config, start it, run a group of tests, stop it again and so on
20:28
<jaysonsantos>
can I for example, run a bash with memcached -p5000& memached -s/tmp/memached.sock ?
20:29
<antares_zzzz>
jaysonsantos: or use env variables (see in the docs) to indicate what part of the suite you want to run
20:29
<antares_zzzz>
jaysonsantos: you will have to use sudo service memcached stop to first stop it but yes
20:30
<jaysonsantos>
ok, then i will try with before_script and after script
20:30
<antares_zzzz>
jaysonsantos: take a look at this guide, it demonstrates a very common technique (that php projects often need but Ruby or JVM languages do not): http://about.travis-ci.org/docs/user/languages/php/, search for "Multiple Versions of Dependencies"
20:31
<antares_zzzz>
jaysonsantos: so you can make your script: to use $ENV_VAR that indicates how exactly you want to start memcached, then you add 3 different rows for it to the env: key in your .travis.yml
20:31
<antares_zzzz>
for 2 python versions this should be 2 * 3 = 6 matrix rows
20:31
<antares_zzzz>
makes sense so far?
20:32
<antares_zzzz>
jaysonsantos: or, you can just write a shell script, add it to ci/run_tests.sh to your repo, make script: run it (just don't forget to chmod +x it) and do all this memcached stopping stuff as part of that script
20:32
<antares_zzzz>
I'd probably do the latter, you can also use the same shell script to run tests this way locally
20:33
<antares_zzzz>
while making complex before_script and after_script work may be difficult
20:33
<jaysonsantos>
antares_zzzz when a test is finished the vm snapshot is dropped ? if yes, i don't have to stop memached stuff
20:33
<antares_zzzz>
jaysonsantos: you don't have to stop it at the very end but you may want to stop it in the middle
20:34
<jaysonsantos>
should this work then?
20:34
<jaysonsantos>
before_script:
20:34
<jaysonsantos>
- "memcached -s/tmp/memcached.sock -d"
20:34
<jaysonsantos>
- "memcached -p5000 -d"
20:35
<antares_zzzz>
yes, both scripts will be run, in order
20:35
<antares_zzzz>
jaysonsantos: or do this: https://gist.github.com/614df272bcc928c07578
20:35
<antares_zzzz>
jaysonsantos: but on boot memcached is already running, you may need to stop it using sudo service memcached stop
20:36
<antares_zzzz>
hm, no, actually, you don't. It is not running on port 5000 by default as far as I know.
20:36
<jaysonsantos>
i will test with default and a non default as 11211 is already running, i don't have to do anything
20:36
<antares_zzzz>
ah, ok
20:37
<jaysonsantos>
antares_zzzz it worked with before_script :D this service is beatiful lol
20:37
<antares_zzzz>
well, the point is, consider using a shell script that you run in script: and you can use sudo or anything you want in that shell script
20:37
<antares_zzzz>
jaysonsantos: glad you like it
20:38
<jaysonsantos>
antares_zzzz i think i will have to stop it because i need to run it with sasl
20:38
<antares_zzzz>
the default on port 11211?
20:39
<jaysonsantos>
antares_zzzz yeah
20:40
<jaysonsantos>
it was with 22 errors now, just 5 :D http://travis-ci.org/#!/jaysonsantos/python-binary-memcached/jobs/755420
20:40
<antares_zzzz>
jaysonsantos: don't stop there :)
20:40
<antares_zzzz>
only 0 failures make us travis maintainers proud!
20:44
<jaysonsantos>
antares_zzzz: i will try with this to check if it works https://gist.github.com/1928437
20:44
<antares_zzzz>
jaysonsantos: are you sure it is memcached stop and not service memcached stop?
20:45
<jaysonsantos>
antares_zzzz: ops, typo
20:45
<antares_zzzz>
we use ubuntu 11.04 at the moment, it is pretty much all converted to upstart
20:46
<antares_zzzz>
jaysonsantos: you need to chmod +x your script
20:47
<antares_zzzz>
by default files are not executable
20:47
<jaysonsantos>
antares_zzzz: done, it is running now
20:47
<jaysonsantos>
humm, server is not built with sasl support
20:48
<jaysonsantos>
antares_ if i'm not wrong, maybe you have to tweak memached install
20:48
<jaysonsantos>
before compiling, installing libsasl-dev if i'm not wrong
20:48
<antares_>
jaysonsantos: we can do it
20:49
<antares_>
jaysonsantos: I can try it. New VM images will be deployed in a day or two.
20:49
<antares_>
we have several other images that need update & deployment as well
20:49
<antares_>
but we do adapt VMs as we get feedback from our users so thank you very much
20:49
<antares_>
we will try to make memcached sasl-aware
20:50
<antares_>
however, we do not compile it from source :/
20:50
<antares_>
we just use regular apt packages
20:50
<antares_>
do you know what we'd have to do in this case?
20:50
<jaysonsantos>
antares_: here is a lillte guide http://code.google.com/p/memcached/wiki/SASLHowto
20:50
<jaysonsantos>
ouch, i don't remember if ubuntu have it just like plugins
20:51
<jaysonsantos>
antares_: maybe i mock auth in this case
20:52
<antares_>
jaysonsantos: please investigate what needs to be done (packages preferred, of course) and we will don it this or next week
20:52
<antares_>
jaysonsantos: for now, probably limiting travis to running only a portion of your tests is the best option
20:53
<jaysonsantos>
antares_: can you guys use ppa?
20:53
<antares_>
jaysonsantos: if it provides packages for 11.04 and 11.10, absolutely
20:53
<antares_>
ah, also, 32-bit
20:53
<antares_>
our VMs are 32-bit
20:54
<antares_>
I found this but it also means compiling from source: http://blog.couchbase.com/sasl-memcached-now-available
20:55
<jaysonsantos>
should be bad do this? 'apt-get build-dep memcached; apt-get install libsasl-dev; apt-get source memached; cd blabla;./configure && make && make install'
20:56
<jaysonsantos>
would it be bad*
20:57
<antares_>
jaysonsantos: we can build memcached from source, we just try hard to first find an option that will allow us to avoid it
20:57
<antares_>
because we already compile quite a bit and things compiled from source tend to be more fragile :(
20:57
<jaysonsantos>
antares_: I know. It sucks.
20:58
<antares_>
tarball URLs break, servers go down, etc
20:58
<antares_>
your idea is nice because we would be using source packages
20:58
<antares_>
which also means we will get the same init scripts, right?
20:58
<antares_>
I really like this idea
20:58
<jaysonsantos>
yeah
20:59
<antares_>
jaysonsantos: we have some immediately issues to resolve (mostly around Python support, actually) but I will try to do it this week
20:59
<jaysonsantos>
and the same version as ubuntu comes with
20:59
<antares_>
jaysonsantos: if you file an issue with travis-ci/travis-cookbooks and assign it to me (@michaelklishin), I definitely won't forget about this
20:59
<antares_>
jaysonsantos: yeah, this is a great idea
21:00
<antares_>
both addresses my concerns and will let us configure memcached the way official guide suggests
21:00
<antares_>
jaysonsantos: do you need a separate user for SASL?
21:00
<jaysonsantos>
antares_: doing it now, thank you!
21:00
<antares_>
I just don't see why a separate user is needed
21:02
<GitHub186>
[travis-ci] jaysonsantos opened issue #434: Build memcached with sasl support. https://github.com/travis-ci/travis-ci/issues/434: http://git.io/PYR7hA
21:03
<jaysonsantos>
antares_: for what?
21:03
<jaysonsantos>
you mean saslpasswd2?
21:08
<antares_>
jaysonsantos: right
21:08
<antares_>
both guides feature creating users :/
21:08
<antares_>
and installing some templates (which is very easy to do with Chef but just extra work)
21:09
<jaysonsantos>
antares_: i think they didn't want to use PAM for example, to work with a lot of SOs
21:10
<antares_>
jaysonsantos: will you mind sticking around in this channel? I will ask questions once we get to adding SASL to memcached.
21:12
<jaysonsantos>
no problem, when do you think it will be?
21:13
<antares_>
jaysonsantos: this week
21:13
<antares_>
jaysonsantos: oops, I will move your ticket to travis-cookbooks
21:13
<antares_>
we are trying to move away from one huge pile of issues for everything
21:13
<antares_>
other than that, I have all the info
21:14
<antares_>
if I can figure out SASL config file + user part (which I am not really familair with), it should not be too hard to put together a new recipe
21:15
<jaysonsantos>
antares_: if i'm not online, contact me on github and I connect on IRC, my profile is @jaysonsantos
21:16
<antares_>
jaysonsantos: sure. Thanks for pointing out that we did not have sasl support.
21:16
<jaysonsantos>
no problem, i've solved some erros on my project and saw that i didn't mocked the auth support too
22:02
<antares_>
dstufft: ping
22:03
<dstufft>
antares_: pong
22:03
<antares_>
dstufft: can you please try 2.5 on travis?
22:03
<antares_>
it should be available now
22:06
<dstufft>
looks to be working
22:07
<dstufft>
my tests will fail
22:07
<dstufft>
but python --verison is workin
22:07
<dstufft>
antares_: stuff is out of order in the build log
22:07
<dstufft>
is that normal
22:08
<antares_>
dstufft: do you have a link?
22:08
<dstufft>
http://travis-ci.org/#!/dstufft/slumber/jobs/755942
22:08
<antares_>
that's not normal but sometimes processes started in the background result in race conditions
22:08
<dstufft>
http://d.stufft.io/0G0J3f0p2i3X2O3N3B0E
22:09
<dstufft>
the first line that it's installing stuff occurs before pip --version output, and before pip install -r is recorded
22:09
<antares_>
dstufft: hm, not sure what is out of order but we just stream stdout and stderr
22:10
<antares_>
dstufft: wait, please refresh the page
22:10
<dstufft>
oh now it's fixed
22:10
<antares_>
dstufft: it may be that messages were delivered to your browser out of order :)
22:10
<antares_>
dstufft: our log collector app (Travis Hub) both updates DB and propagates them via Web Sockets to browsers
22:10
<antares_>
in the 2nd case, we rely on pusher to deliver them properly
22:11
<antares_>
DB logs should always be in order because we carefully design Hub to not have race conditions (as far as messages from one specific build go)
22:11
<antares_>
dstufft: makes sense?
22:11
<antares_>
dstufft: so 2.5 is good now?
22:11
<dstufft>
looks like it
22:11
<antares_>
ok, I will add it back to the docs
22:12
<dstufft>
antares_: it would be cool if you added a requirements thing similar to Gemfile
22:12
<dstufft>
where you can specify multiple requirements files to test different versions
22:12
<dstufft>
in the matrix
22:12
<antares_>
to force projects to have it?
22:13
<antares_>
yeah that shouldn't be hard
22:13
<dstufft>
nah don't force it, just if they are using the default script or whatever
22:13
<antares_>
permutations are trivial to add
22:13
<dstufft>
I mean they can do it with env variables, but the gemfiles thing is a little cleaner Ithink
22:14
<antares_>
yes but it will require good docs on this
22:14
<antares_>
Gemfile/--gemfile is something bundler supported for a long time
22:14
<antares_>
a ton of people know about it
22:14
<antares_>
here we will have to introduce a convention
22:14
<antares_>
but it's doable and of course it makes sense to try
22:18
<GitHub53>
[travis-ci.github.com] michaelklishin pushed 1 new commit to master: http://git.io/DMDw4Q
22:18
<GitHub53>
[travis-ci.github.com/master] Revert "We documented Python 2.5 availability too soon. See https://github.com/travis-ci/travis-cookbooks/issues/29." - Michael S. Klishin
22:20
<dstufft>
antares_: the "examples" section has fabric/fabric linked too http://travis-ci.org/#!/fabric/fabric/blob/master/.travis.yml
22:20
<dstufft>
instead of github
22:22
<GitHub5>
[travis-ci.github.com] michaelklishin pushed 1 new commit to master: http://git.io/JhB5UA
22:22
<GitHub5>
[travis-ci.github.com/master] Correct fabric example link - Michael S. Klishin
22:38
<dreid>
antares_: Can you also change the klein link to point to the official repository? (Which is oddly not the root of the forks) https://github.com/MostAwesomeDude/klein/blob/master/.travis.yml
22:39
<antares_>
dreid: sure
22:39
<dreid>
antares_: Thanks.
22:39
<kennethreitz>
antares_: i think an small addition to the default install may be in order
22:40
<kennethreitz>
antares_: - `pip install . --use-mirrors` + `pip install -r requirements.txt --use-mirrors`
22:40
<antares_>
kennethreitz: ok, we will do it some time this week
22:40
<zupo>
I'm trying out Travis CI for Plone based project
22:40
<GitHub196>
[travis-ci.github.com] michaelklishin pushed 1 new commit to master: http://git.io/FYCTfg
22:40
<GitHub196>
[travis-ci.github.com/master] Update klein example link to point to the new official repo - Michael S. Klishin
22:41
<kennethreitz>
antares_: thanks man :) should i send a pull request?
22:41
<zupo>
Plone consists of +200 packages and downloading them all takes quite some time
22:41
<zupo>
is there a way I can cache downloads, or make a custom image snapshot with all packages already downloaded?
22:41
<zupo>
or extend the time limit on my builds
22:42
<antares_>
kennethreitz: if you will update tests, sure :) https://github.com/travis-ci/travis-build/blob/master/lib/travis/build/job/test/python.rb#L23, https://github.com/travis-ci/travis-build/blob/master/spec/build/job/test/python_spec.rb#L22
22:42
<kennethreitz>
antares_: you got it :)
22:42
<antares_>
kennethreitz: tests use mocks and I am afraid it will be quite painful for you to quickly update. I can do it later this week.
22:42
<kennethreitz>
antares_: the tests are wrong now actually
22:42
<antares_>
zupo: we can consider preinstalling plone
22:43
<kennethreitz>
antares_: requirements shouldn't be captilized
22:43
<antares_>
kennethreitz: we try both
22:43
<kennethreitz>
oh, weird
22:43
<zupo>
antares_: won't help
22:43
<antares_>
zupo: why?
22:43
<zupo>
whatever is in Ubuntus repos is ages old
22:43
<kennethreitz>
zope isn't really possible
22:43
<antares_>
zupo: we are not tied to ubuntu repos
22:43
<kennethreitz>
you can't run it on heroku either
22:43
<antares_>
zupo: can this stuff be installed via pip?
22:43
<kennethreitz>
it takes over 30 minutes to istall
22:43
<kennethreitz>
*install
22:43
<kennethreitz>
it's the worst
22:44
<iElectric>
not that much
22:44
<dstufft>
kennethreitz: antares_ make sure you only do pip install -e . if setup.py exists
22:44
<zupo>
antares_: tricky, but I've heard of people pulling it off
22:44
<antares_>
zupo: we may consider bumping timeouts a bit but it sounds like the only realistic way is to preinstall plone
22:44
<dreid>
kennethreitz: Wow… I do not feel so bad about Twisted now.
22:44
<zupo>
I'm trying to minimize my build now, to get only what is *really* needed
22:44
<zupo>
will update you guys
22:44
<kennethreitz>
it's not really designed for pip, they use buildbot
22:44
<antares_>
dstufft, kennethreitz: I will forget this all and come back to ask you what we should do once we get to it :)
22:44
<kennethreitz>
which is a whole other thing
22:44
<dstufft>
no problem :)
22:44
<zupo>
kennethreitz: buildout, not buildbot
22:44
<zupo>
2 different things
22:45
<kennethreitz>
that's what i meant sorry :)
22:45
<kennethreitz>
yeah i know
22:45
<kennethreitz>
haha
22:45
<dstufft>
buildout would work tehcincally
22:45
<dreid>
buildbot is a whole other thing too. :)
22:45
<iElectric>
it does
22:45
<dstufft>
you'd just have to provide your own commands
22:45
<dstufft>
instead of the default ones
22:45
<kennethreitz>
buildout takes more than 10 minutes
22:45
<* dreid>
considers trying to build buildbot with travis.
22:45
<kennethreitz>
i have a buildout buildpack for heroku
22:45
<antares_>
zupo: basically, we can consider preinstalling heavy popular packages as part of VM provisioning but if something takes 30 minutes to install, it is out of question even for this case
22:45
<iElectric>
if you has ssd
22:45
<iElectric>
it would be under 10min
22:45
<kennethreitz>
but i've never run it successfully, because nothing takes less than 10 minutes
22:45
<antares_>
our entire PPP VM image is built in maybe 15 minutes
22:46
<zupo>
understandable
22:46
<antares_>
and it has a ton of stuff and 9 (!) runtimes is compiled from source
22:46
<zupo>
my build is currently 12 mins on my local machine
22:46
<iElectric>
antares_: it's just network io
22:46
<zupo>
on a shaky network
22:46
<iElectric>
that you need to install few hundred packages
22:46
<kennethreitz>
NO BIG DEAL
22:46
<zupo>
but yeah, like iElectric says, it's mostly due to network io
22:46
<kennethreitz>
zope should not be a target
22:47
<dstufft>
a few hundred packages :|
22:47
<kennethreitz>
it's old deprecated software
22:47
<antares_>
iElectric, zupo: I understand but we cannot add 30 minutes to VM provisioning
22:47
<zupo>
kennethreitz: :P
22:47
<kennethreitz>
that exists only because ANCIENT systems are still around
22:47
<zupo>
antares_: sure, I'm not asking you to
22:47
<kennethreitz>
but consulting $$
22:47
<antares_>
we sometimes need to quickly rebuild them and deploy a fix, etc
22:47
<antares_>
like with python2.5 today
22:47
<zupo>
I'm trying to find a way to squeeze under 10mins, that's all
22:48
<antares_>
zupo: I can check but I think for machines where python VMs run the timeout is 10-12 minutes
22:48
<antares_>
zupo: you can cheat by splitting your installation into multiple commands (install: can take an array of lines)
22:48
<dstufft>
zupo: just saying, 12 minutes is crazy for an install ;P that would drive me batty
22:49
<antares_>
zupo: but of course if some project takes out a worker for 15-25 minutes it is not appreciated
22:49
<iElectric>
it's not like you install software everyday
22:49
<zupo>
dstufft: it's not 12, expect you do a fresh install
22:49
<dstufft>
I thought my 2-3 minutes installs were bad
22:49
<antares_>
we have very severe cases when huge Ruby on Rails apps like diaspora do that and people often complain
22:49
<zupo>
if you have a local cache of all packages it's under a minute
22:49
<dstufft>
*does a fresh install on every deploy*
22:49
<kennethreitz>
dstufft: you crazy
22:49
<dstufft>
kennethreitz: I am slightly ;)
22:50
<iElectric>
dstufft: we have this software
22:50
<kennethreitz>
dstufft: crazy enough to something about it - i like it :)
22:50
<iElectric>
that tries to do siffs
22:50
<iElectric>
so you don't have to redeploy everything
22:50
<antares_>
dstufft: no, 2-3 minutes is probably just slightly over average for Ruby projects (with bundler 1.1 it will be a lot faster because it is a lot more efficient in # of network round trips)
22:51
<dstufft>
antares_: within a week i'm going to release a new endpoint on crate that should be a lot faster for pip… but not every package will install from it :(
22:51
<dstufft>
the packages uploaded by assholes won't work
22:51
<antares_>
zupo: so, we will see what our options are but I am afraid it may be too heavy for travis
22:51
<antares_>
dstufft: how do you detect packages uploaded by assholes?
22:52
<zupo>
totally understandable
22:52
<dstufft>
antares_: easy, they arn't hosting their packages and PyPI and instead is forcing pip to spider a shitton of external links
22:52
<dstufft>
packages on PyPI*
22:52
<dstufft>
so i'm going to make an endpoint on crate that just removes everything but the actual package downloads
22:52
<antares_>
dstufft: that sounds really weird, so they basically just host .egg files or something?
22:53
<iElectric>
antares_: what if you only limit whole build to 20min?
22:53
<iElectric>
that would make more sense
22:53
<antares_>
iElectric: we can do that but so far we limit each command and it works very well in general
22:53
<zupo>
antares_: a dir that is shared among builds totally solves the Plone problem
22:53
<antares_>
iElectric: the point is that you don't want to wait 20 minutes if someone committed a line that hangs
22:54
<antares_>
zupo: no, we don't keep shared state
22:54
<zupo>
download all packages once, use them later for subsequent builds
22:54
<iElectric>
waiting is pretty cheap
22:54
<zupo>
antares_: yeah, I know, just saying
22:54
<antares_>
zupo: shared state is what you don't want in general. The only reason why travis is so flexible and you have sudo is because we snapshot VMs
22:54
<dstufft>
antares_: basically PyPI started out as just an index, you couldn't host anything on it. it was just a link of urls that you could go to find packages, then automated tools came out that would spider those and download packages (easy_install), then PyPI allowed uploads and started hosting packages, but still left people use a link instead of a hosting on PyPI, somtime after that pip was created, but in order to maintain compatab
22:54
<dstufft>
had to ontinue spidering things
22:54
<antares_>
zupo: what you are suggesting is basically preinstalling
22:55
<zupo>
antares_: kind of
22:55
<dstufft>
antares_: example: https://simple.crate.io/PIL/ PIL isn't available on PyPI, you have to go to the pages linked there in order to find the download you want to install
22:55
<antares_>
dstufft: I had no idea how awesome maven-based tools and rubygems/rubygems.org are until we started supporting python :)
22:55
<kennethreitz>
hahahahaha
22:56
<dreid>
rebar is the way to go.
22:56
<dreid>
My requirements.txt are made up mostly of github urls. :)
22:56
<antares_>
dreid: rebar is also nice but it does not really solve package distribution problem. A friend of mine did agner which was only half successful.
22:57
<antares_>
my favorite tool of all is Leiningen (Clojure) but it is Maven-based and only adds very easy to use jar hosting (clojars.org) + can use any other Maven repo
22:58
<dstufft>
antares_: here's a really bad example ;P
22:58
<dstufft>
https://simple.crate.io/Plone/
22:58
<dstufft>
that is uh
22:58
<antares_>
it can also use local source checkouts (when you are working on a lib as part of some app) and has a bunch of other awesome features
22:58
<dstufft>
218 links
22:58
<dstufft>
that pip has to spider
22:58
<dstufft>
now you don't normally install Plone from pip
22:58
<dstufft>
but there's other ones like that
22:58
<antares_>
dstufft, kennethreitz, dreid: did we guys tell you why SSH sessions fail sometimes?
22:58
<antares_>
with Python projects and almost never with any others?
22:59
<dstufft>
was it because pip opens up a ton of connections to try and speed up spidering a billion links?
22:59
<antares_>
pip opens a crazy # of network connections and it overwhelms our VM NICs which use NAT
22:59
<antares_>
so NIC starts dropping connections and eventually drops ssh :)
22:59
<iElectric>
over 65k?
22:59
<antares_>
it uses LRU or something
22:59
<dreid>
antares_: Yes. :(
22:59
<antares_>
iElectric: hundreds from what we can tell so far
22:59
<kennethreitz>
hahahaha
22:59
<antares_>
fortunately we found a workaround
22:59
<kennethreitz>
thats amazing
23:00
<iElectric>
you can do two levels of nat
23:00
<kennethreitz>
antares_: what was it?
23:00
<dstufft>
antares_: you won't be able to default to my new crate endpoint, but you can suggest it if their package is compatable
23:00
<iElectric>
great firewall of china
23:00
<dstufft>
it should speed things up *alot* for some people
23:00
<antares_>
but I was shocked to realize that virtual NICs are overloaded by 1 tool
23:00
<antares_>
kennethreitz: we use different NIC type, it somehow can handle more connections
23:00
<antares_>
the one virtualbox uses by default cannot
23:01
<antares_>
dstufft: how can we suggest your crate endpoint?
23:02
<antares_>
I read something on pip yesterday trying to understand if we can throttle it
23:02
<antares_>
I saw mirror lists
23:02
<antares_>
we will add crate to it?
23:03
<dstufft>
antares_: crate will hopefully be an official mirror soon
23:03
<iElectric>
but seriously, pip creating hundreds of connections
23:03
<dstufft>
I'm working out some of the kinks
23:03
<iElectric>
that shouldn't happen
23:04
<zupo>
antares_: slimmed it down to 8 minutes locally
23:04
<zupo>
trying again on travis
23:05
<dstufft>
antares_: once it's done just mentioning in travis docs as a possible way to speed up your test installs would probably be fine
23:05
<antares_>
iElectric: in the Ruby community a lot of people bitch about how bundler is slow and all. It does maybe a few dozens of roundtrips even for Gemfiles with 15-20 gems. Bundler 1.1 is a lot more efficient in this regard (3-4 roundtrips for 15-20 gems is quite common) but even then people don't like it.
23:05
<dstufft>
they can just add -i https://mynewendpoint.crate.io/ to their requirements file, or to the install command
23:05
<iElectric>
antares_: roundstrip meaning redirects?
23:05
<antares_>
iElectric: request/response
23:06
<dstufft>
pip is doing better than the alternatives, this is broken at the PyPI level
23:06
<zupo>
in any case … this build times just encourage me to do more pyramid projects (instead of plone) :)
23:06
<antares_>
iElectric: it fetches gems metadata when it is calculating final dependency list (with exact versions)
23:06
<dstufft>
antares_: I actually plan on doing a *real* api on crate, instead of spidering links, and stealing how bundler does it's gem version resoltution
23:06
<dstufft>
it's really nice
23:06
<antares_>
dstufft: I understand. Bundler 1.1 is much more efficient because of rubygems.org API improvements added specifically for bundler.
23:07
<dstufft>
i won't be able to do it in pip because it'll be crate specific
23:07
<antares_>
dstufft: and is really complex. But yes, Ruby community more or less solved it, so it is a good idea to start with what bundler does.
23:07
<dstufft>
the PyPI guys are pretty any change
23:07
<iElectric>
s/guys/guy/
23:07
<antares_>
don't they suffer from the same problem?
23:07
<antares_>
I mean, as Python developers?
23:08
<dstufft>
I don't know honestly how they put up with it
23:08
<iElectric>
antares_: if someone would sit down and solve it
23:08
<iElectric>
I'm sure they would accept the patch
23:08
<dstufft>
no they wouldn't
23:08
<antares_>
this is sad
23:08
<dstufft>
they flat out told me no to adding OAuth, I offered to write it
23:08
<dstufft>
I convinced them
23:09
<iElectric>
but that's something different
23:09
<iElectric>
something is improving performance
23:09
<iElectric>
and another thing is adding additional authentication services
23:09
<dstufft>
right, but you can't improve performance without making a new api
23:10
<dstufft>
the issues are a cause of the current API
23:10
<dstufft>
they will not accept a patch that does not allow people to continue to host things on their own servers
23:10
<iElectric>
you would have to do it for distutils2
23:10
<iElectric>
and wait couple of years :-)
23:10
<dstufft>
which is where *a lot* of the slow down comes from
23:10
<iElectric>
or get Tarek drunk
23:10
<iElectric>
which is not hard
23:10
<dstufft>
iElectric: or I could do it myself, add it to crate.io and say fuck em
23:11
<dstufft>
I gave up on trying to improve PyPI, I've got better things to do then argue on catalog-sig
23:11
<dstufft>
I just mirror PyPI and do shit myself now
23:11
<iElectric>
catalog-sig is worst toilet of internet than 4chan
23:12
<dstufft>
MvL basically told me *no* for switching to sha256
23:12
<dstufft>
instead of md5
23:12
<dstufft>
which md5 is broken
23:12
<iElectric>
for package hashe?
23:12
<antares_>
actually, bundler originally has happened for two reasons
23:13
<dstufft>
iElectric: yup
23:13
<iElectric>
that makes sense
23:13
<iElectric>
you dont need sha
23:13
<dstufft>
iElectric: you do know that I can contruct 2 packages, both hash to the same thing, one is useful, one it malevolant
23:13
<antares_>
1. People were fed up with rubygems state of affairs, 2. Large Rails apps were no longer uncommon and everyone learned how hard it is to managed dependencies and isolate environments.
23:13
<dstufft>
this is not a theory
23:13
<dstufft>
this is doable _today_
23:13
<iElectric>
dstufft: but you can't upload them to my package name
23:13
<antares_>
so bundler, while it uses rubygems under the hood, was an FU to rubygems maintainers back in the day (or so I think)
23:13
<iElectric>
I know.
23:14
<dstufft>
iElectric: no, but your credentials are sent plaintext to PyPI
23:14
<iElectric>
if you don't use vpn everywhere except home/work
23:14
<iElectric>
you must be crazy
23:14
<antares_>
ok folks, it is time for me to take a nap. It is past 8 a.m. here in Moscow :(
23:14
<iElectric>
that's why I wrote this
23:14
<iElectric>
https://gist.github.com/1547663
23:15
<antares_>
good night and thanks for the awesome Python launch
23:15
<antares_>
over 110 projects in 1 day
23:15
<iElectric>
thanks for killing my 2 month fight with jenkins
23:35
<zupo>
antares_zzzz: FYI: I managed to trim it down to 5 mins
23:35
<zupo>
now you have proof that you can CI even such a mastodon as Plone is on Travis
23:35
<zupo>
http://travis-ci.org/#!/niteoweb/niteoweb.click2sell/builds/756318
23:58
<GitHub129>
[travis-build] michaelklishin pushed 3 new commits to master: http://git.io/-hq06Q
23:58
<GitHub129>
[travis-build/master] -vv is too verbose, -q has similar output to gem
23:58
<GitHub129>
[travis-build/master] use --quiet on perl spec as well - Lee Aylward
23:58
<GitHub129>
[travis-build/master] Merge pull request #6 from leedo/patch-1 - Michael Klishin