/irc-logs / w3c / #webapps / 2013-04-15 / end
Options:
- # Session Start: Mon Apr 15 00:00:01 2013
- # Session Ident: #webapps
- # [00:21] * Quits: glenn (~gadams@public.cloak) (Client closed connection)
- # [00:31] * Joins: chaals (~Adium@public.cloak)
- # [00:31] * Quits: chaals1 (~Adium@public.cloak) (Client closed connection)
- # [00:34] * Joins: glenn (~gadams@public.cloak)
- # [01:17] * heycam|away is now known as heycam
- # [02:36] * Joins: shepazu (schepers@public.cloak)
- # [03:04] * Joins: glenn_ (~gadams@public.cloak)
- # [03:04] * Quits: glenn (~gadams@public.cloak) (Client closed connection)
- # [03:15] * Quits: glenn_ (~gadams@public.cloak) (Client closed connection)
- # [03:15] * Joins: glenn (~gadams@public.cloak)
- # [03:30] * Quits: glenn (~gadams@public.cloak) (Client closed connection)
- # [04:07] * heycam is now known as heycam|away
- # [04:27] * Joins: glenn (~gadams@public.cloak)
- # [04:41] * heycam|away is now known as heycam
- # [05:58] * Quits: glenn (~gadams@public.cloak) (Client closed connection)
- # [07:08] * heycam is now known as heycam|away
- # [07:39] * heycam|away is now known as heycam
- # [08:07] * Joins: richt (~richt@public.cloak)
- # [08:08] * Joins: glenn (~gadams@public.cloak)
- # [09:04] * Joins: dom (dom@public.cloak)
- # [09:29] * Joins: darobin (rberjon@public.cloak)
- # [09:42] * Joins: Ms2ger (~Ms2ger@public.cloak)
- # [10:10] * Joins: Dashiva (~noone@public.cloak)
- # [10:16] * Quits: dom (dom@public.cloak) (Ping timeout: 60 seconds)
- # [10:53] * Joins: dom (dom@public.cloak)
- # [10:56] * heycam is now known as heycam|away
- # [11:33] * Quits: richt (~richt@public.cloak) (Client closed connection)
- # [11:33] * Joins: richt (~richt@public.cloak)
- # [11:35] * heycam|away is now known as heycam
- # [11:37] * Quits: richt (~richt@public.cloak) (Ping timeout: 60 seconds)
- # [11:55] * Quits: Ms2ger (~Ms2ger@public.cloak) ("bbl")
- # [12:05] * Joins: abarsto (~abarsto@public.cloak)
- # [12:05] * abarsto is now known as ArtB
- # [12:21] * Joins: richt (~richt@public.cloak)
- # [12:23] * heycam is now known as heycam|away
- # [12:56] * Joins: Ms2ger (~Ms2ger@public.cloak)
- # [14:12] * Quits: Ms2ger (~Ms2ger@public.cloak) (Ping timeout: 60 seconds)
- # [14:14] * Joins: richt_ (~richt@public.cloak)
- # [14:16] * Joins: Ms2ger (~Ms2ger@public.cloak)
- # [14:17] * Quits: richt (~richt@public.cloak) (Ping timeout: 60 seconds)
- # [14:18] * Quits: richt_ (~richt@public.cloak) (Ping timeout: 60 seconds)
- # [14:29] * Joins: richt (~richt@public.cloak)
- # [14:36] <ArtB> FYI, I just added some new subitems for WebIDL, Testing and AppCache http://www.w3.org/wiki/Webapps/April2013Meeting#Potential_Topics
- # [14:41] <darobin> ArtB: did you see what http://www.w3.org/mid/516BC214.7020600@w3.org ?
- # [14:41] <darobin> I would feel more comfortable if you'd weigh in
- # [14:42] <darobin> rather than just inviting people to attend your group's meetings without asking you :)
- # [14:46] * Quits: shepazu (schepers@public.cloak) ("is sleepy")
- # [14:46] <ArtB> darobin, I'm totally fine with this but I'm not responsible for the hosting. It could be too late.
- # [14:47] <ArtB> (for the April 25-26 meeting)
- # [14:48] <darobin> ArtB: yeah, I wasn't so much thinking about logistics as much as on the principle
- # [14:48] <darobin> I think that it would be extremely useful to increase communication with TC39
- # [14:48] <ArtB> makes sense to me
- # [14:48] <darobin> and giving them a little bit of time in WebApps would go a long way
- # [14:48] <ArtB> ok
- # [14:48] <darobin> note that logistics wise, I think that WebApps already has TC39ers attending, so it could just be a question of scheduling it
- # [14:48] <ArtB> (I'm not sure about the bit-o-time be real useful but it's certainly worth a try)
- # [14:49] <ArtB> ok, that'd be great. Do you have any specific names?
- # [14:50] <Ms2ger> darobin, as long as you check they don't bring rotten tomatoes into meetings about WebIDL
- # [14:50] <darobin> Ms2ger: why would I want to check that? so long as no one tells them that starting work on that spec was my idea we're fine aren't we? O:-)
- # [14:51] * Ms2ger accidentally lets that slip on es-discuss
- # [14:51] <darobin> ArtB: no, I don't have the db — I was thinking that slightlyoff might be there but apparently not
- # [14:51] <darobin> I'll see if I can dig something up
- # [14:54] * Joins: richt_ (~richt@public.cloak)
- # [14:57] * Quits: richt (~richt@public.cloak) (Ping timeout: 60 seconds)
- # [15:01] * Joins: smaug (~chatzilla@public.cloak)
- # [15:09] * Quits: dom (dom@public.cloak) (Ping timeout: 60 seconds)
- # [15:13] * Joins: davidb (~davidb@public.cloak)
- # [15:26] * Joins: shepazu (schepers@public.cloak)
- # [15:26] <Yves> perfect, thanks!
- # [15:31] * Joins: dom (dom@public.cloak)
- # [16:06] <dom> jgraham, I'm not sure I get how I am supposed to plug github notifications into your script
- # [16:06] <dom> (which otherwise is already working great, cf https://w3c-test.org/web-platform-tests/submissions/ )
- # [16:08] <jgraham> dom: You need to register a webhook
- # [16:08] <jgraham> Using the github api
- # [16:08] <darobin> can't you just register in the UI?
- # [16:08] <dom> but that requires exposing your script as a CGI, right?
- # [16:08] <jgraham> to get push, pull_request and issue_comment events
- # [16:08] <jgraham> Yes, somewhere the script has to be exposed
- # [16:09] <jgraham> Not necessarily on w3c-test.org of course
- # [16:09] <darobin> we can make that somewhat recursive and have the script be a checkout that's updated automatically :)
- # [16:10] <dom> jgraham, any chance I could make you make it react to email notifs instead?
- # [16:10] <jgraham> You W3C people love your email notifications, don't you :p
- # [16:10] <darobin> heh
- # [16:11] <darobin> dom: I don't know if you can get email for all of those events
- # [16:11] <dom> well, email has reasonably good fail over mechanisms
- # [16:11] * Ms2ger remembers how much fun email caused for tinderbox at Mozilla
- # [16:11] <jgraham> Sure, I guess it is a more reliable message channel. But I don't know if you can actually get all the data you need
- # [16:12] <jgraham> and programming against email messages doesn't seem much like fun
- # [16:12] <dom> yeah :/
- # [16:12] <darobin> yeah, I wouldn't bet on that
- # [16:12] <darobin> the email hooks do send you some amount of data, but you have to parse it out of the message
- # [16:12] <jgraham> Maybe more companies should expose their APIs over SMTP rather than HTTP
- # [16:12] <darobin> and I'm not sure they promise it won't change
- # [16:12] * Joins: richt (~richt@public.cloak)
- # [16:12] <darobin> I also don't think that the email hook gets things like comments and new issues
- # [16:13] <darobin> I reckon it only gets pushes
- # [16:13] * dom wonders if there are webhook-to-smtp systems
- # [16:13] <darobin> dom: is it a big deal to set it up CGI?
- # [16:14] <darobin> using that "web" thing and all
- # [16:14] <dom> probably not, but having high confidence on its reliability would likely be
- # [16:14] <dom> and given the history of w3c-test.org in that regard, I'm not too enthused
- # [16:14] <jgraham> You don't really need high confidence in its reliability
- # [16:15] <darobin> dom: this isn't a life-or-death application
- # [16:15] <darobin> I say ship it and if it's too unreliable then we'll see
- # [16:15] * Quits: richt_ (~richt@public.cloak) (Ping timeout: 60 seconds)
- # [16:15] <jgraham> One helpful enhancement would probably be to make just calling sync.py with no arguments try to update all pull requests that already exist
- # [16:16] <jgraham> Then you have an easy way to recover many types of missed events
- # [16:16] <dom> that would help for sure
- # [16:16] * Quits: richt (~richt@public.cloak) (Ping timeout: 60 seconds)
- # [16:16] <darobin> note that even for the pull things that rely on email I still set up a cron job once a day in case something drops
- # [16:17] <darobin> at some point we could do the same here
- # [16:17] <dom> hmm... jgraham, your script currently assumes that what is posted is always true
- # [16:17] <jgraham> I don't understand?
- # [16:17] <dom> yeah, sorry I'm not being very clear
- # [16:18] <dom> what gets POSTed is assumed to be coming from github necessarily
- # [16:18] <dom> and thus is trusted
- # [16:18] <dom> so one could POST that there pull request is worth mirroring
- # [16:18] <dom> if they know the end point URL
- # [16:18] <jgraham> That's a reasonable point
- # [16:19] <darobin> "The Public IP addresses for these hooks are: 207.97.227.253/32, 50.57.128.197/32, 108.171.174.178/32, 50.57.231.61/32, 204.232.175.64/27, 192.30.252.0/22."
- # [16:19] <darobin> so you could check at least that
- # [16:19] <darobin> you could also add a secret as part of the URL
- # [16:19] <dom> arguably that could be left to the CGI deployment layer
- # [16:20] <dom> secret URLs, brrr
- # [16:20] <darobin> dom: well, there are two layers here
- # [16:20] <darobin> there's finding the thing, and then there's passing extra info
- # [16:20] <darobin> you could tell the GH hook to post to /path/to/script?sikkrit=RobinIsDaft
- # [16:21] <darobin> that makes it harder to guess
- # [16:21] <darobin> add to that blocking to the aforementioned IPs and I think you're making it harder to break than socially engineering us would be
- # [16:21] <jgraham> Could use PubSubHubbub
- # [16:21] <darobin> (note that that's how the email hook is protected, it uses a shared secret)
- # [16:21] <jgraham> That seems to have some kind of origin verification built in
- # [16:22] <dom> I was thinking more on the line of using the hook purely as the notification, and then use https to fetch the actual data
- # [16:23] <dom> re-using a reasonably well-tested way to ensure we're getting data from who we think we're getting data from :)
- # [16:23] <darobin> that makes sense too
- # [16:24] <jgraham> Well that seems quite tedious
- # [16:24] <darobin> also, if it's something that can work pure-pull then it means we can apply the same daily cron fallback trick
- # [16:24] <darobin> jgraham: the idea is to have a computer do it
- # [16:24] <jgraham> Tedious to script I mean
- # [16:24] * darobin is Mr Clarity
- # [16:24] <jgraham> Since you are given all the data and then you have to figure out a URL to grab all the data from again
- # [16:25] <dom> they don't give you this in the webhook? what a pity
- # [16:25] <jgraham> I guess they might give you the URL
- # [16:26] <dom> doesn't look like it https://help.github.com/articles/post-receive-hooks :(
- # [16:26] <darobin> mmmm, but that's just for pushes
- # [16:28] <darobin> oh, right, there are no other notifications in the hook system — I thought there were more
- # [16:28] <jgraham> I think something like data["pull_request"]["url"] shoudl give the API URL (from memory)
- # [16:29] <darobin> so, how about we do this differently...
- # [16:29] <darobin> we could use a... manifest!
- # [16:29] <darobin> on push, check the manifest (URL known), and grab the list of PRs to deploy from it
- # [16:30] <darobin> since changing the manifest to add a PR requires a push, the notifications are the right ones
- # [16:30] <darobin> and write-access is okay
- # [16:30] <darobin> I mean it's checke
- # [16:30] <darobin> checked
- # [16:31] <darobin> it's a bit more of a hurdle for us, but not huge — the modifications can be done online by editing in the GH UI
- # [16:31] <darobin> *and* this can work as an email hook
- # [16:31] * Ms2ger wonders what darobin is on about
- # [16:31] <darobin> s/ about// ?
- # [16:32] <jgraham> That sounds like a rewrite :(
- # [16:32] <Ms2ger> darobin, French cheese?
- # [16:32] <jgraham> Not that I mind if it's clearly a better solution, but I don't think it is
- # [16:32] <darobin> I don't know man, I'm just trying to look for a solution...
- # [16:32] <darobin> I can't say I like it either
- # [16:32] <darobin> Ms2ger: why just French?
- # [16:32] <dom> I think we'll go for the webhook and try to secure it as much as we can
- # [16:32] <jgraham> I have never used PubSubHubbub, but it looks like it solves this for us
- # [16:33] <dom> and then we can switch to pubsubhubbub once someone manages to type it without typo
- # [16:33] <jgraham> So using that would be my first instinct
- # [16:33] <darobin> I am leery of solutions involving names I can't type
- # [16:33] <Ms2ger> pshb
- # [16:33] <darobin> that WFM though, was just looking at alternatives
- # [16:35] <jgraham> It is a really very badly named idea
- # [16:43] * Quits: davidb (~davidb@public.cloak) (Ping timeout: 60 seconds)
- # [16:48] <darobin> yeah you can smell the stale beer just on the name
- # [16:53] * Joins: richt (~richt@public.cloak)
- # [16:55] * Joins: richt_ (~richt@public.cloak)
- # [16:55] * Quits: richt (~richt@public.cloak) (Client closed connection)
- # [16:55] * Joins: richt (~richt@public.cloak)
- # [16:59] * Quits: richt_ (~richt@public.cloak) (Ping timeout: 60 seconds)
- # [17:06] * Quits: Ms2ger (~Ms2ger@public.cloak) ("bbl")
- # [17:13] * Joins: davidb (~davidb@public.cloak)
- # [17:58] * Quits: dom (dom@public.cloak) (Client closed connection)
- # [17:58] * Quits: richt (~richt@public.cloak) (Client closed connection)
- # [17:58] * Joins: richt (~richt@public.cloak)
- # [18:02] * Quits: richt (~richt@public.cloak) (Ping timeout: 60 seconds)
- # [18:05] * Joins: sicking (~sicking@public.cloak)
- # [18:23] * Quits: darobin (rberjon@public.cloak) (Client closed connection)
- # [18:36] * Joins: jsbell (~jsbell@public.cloak)
- # [18:38] * Quits: heath (~quassel@public.cloak) ("No Ping reply in 180 seconds.")
- # [18:38] * Joins: heath (~quassel@public.cloak)
- # [18:47] * Quits: heath (~quassel@public.cloak) ("No Ping reply in 180 seconds.")
- # [18:48] * Joins: heath (~quassel@public.cloak)
- # [18:51] * Joins: jeffh (~4441a944@public.cloak)
- # [18:53] * Quits: sicking (~sicking@public.cloak) (Ping timeout: 60 seconds)
- # [18:58] * Joins: dgrogan (~dgrogan@public.cloak)
- # [18:58] * Quits: shepazu (schepers@public.cloak) ("is sleepy")
- # [19:02] * Quits: jeffh (~4441a944@public.cloak) ("http://www.mibbit.com ajax IRC Client")
- # [19:27] * Joins: jeffh (~4441a944@public.cloak)
- # [19:27] * Joins: shepazu (schepers@public.cloak)
- # [19:34] * Quits: jeffh (~4441a944@public.cloak) ("http://www.mibbit.com ajax IRC Client")
- # [19:50] * Joins: jeffh (~4441a944@public.cloak)
- # [19:54] * Joins: davidb_ (~davidb@public.cloak)
- # [19:57] * Quits: davidb (~davidb@public.cloak) (Ping timeout: 60 seconds)
- # [19:57] * davidb_ is now known as davidb
- # [20:22] * Joins: Lachy (~Lachy@public.cloak)
- # [20:23] * Quits: davidb (~davidb@public.cloak) (Ping timeout: 60 seconds)
- # [20:44] * Quits: jeffh (~4441a944@public.cloak) ("http://www.mibbit.com ajax IRC Client")
- # [21:24] * Joins: darobin (rberjon@public.cloak)
- # [21:25] * Joins: davidb (~davidb@public.cloak)
- # [21:34] * Joins: tantek (~tantek@public.cloak)
- # [21:55] * Quits: ArtB (~abarsto@public.cloak) ("Leaving.")
- # [21:57] * Joins: abarsto (~abarsto@public.cloak)
- # [21:57] * abarsto is now known as ArtB
- # [21:59] * Joins: davidb_ (~davidb@public.cloak)
- # [22:00] * Quits: davidb (~davidb@public.cloak) (Ping timeout: 60 seconds)
- # [22:00] * davidb_ is now known as davidb
- # [22:17] * Joins: jeffh (~4441a944@public.cloak)
- # [22:34] * Quits: davidb (~davidb@public.cloak) (Ping timeout: 60 seconds)
- # [22:48] * Quits: jgraham (~jgraham@public.cloak) ("Changing server")
- # [22:53] * Joins: jgraham (~jgraham@public.cloak)
- # [23:11] * Quits: shepazu (schepers@public.cloak) ("is sleepy")
- # [23:28] * Quits: jeffh (~4441a944@public.cloak) ("http://www.mibbit.com ajax IRC Client")
- # [23:40] * Joins: dgrogan_ (~dgrogan@public.cloak)
- # [23:40] * Quits: dgrogan_ (~dgrogan@public.cloak) (Client closed connection)
- # Session Close: Tue Apr 16 00:00:00 2013
The end :)