mozilla :: #conduit

14 Mar 2017
03:26imadueme|afkdavidwalsh, ckolos, smacleod: My bad, the config example I handed off has a bug. The CSP connect-src should be to the API host, not the UI host. So in the example it should read "connect-src;" not "connect-src;" and in this case it should be ""
03:27imadueme|afkalso what davidwalsh said before is correct - about https://
13:12mcoteimadueme|afk: is that the entirety of the problem?
14:08imadueme|afkmcote: yup
14:08mcotemaybe you want to shoot that in an email to ckolos, just to make sure
14:09imaduemewill do
15:04marsdavidwalsh: good morning! I'm hanging out in sococo, looking up how mercurial server-side extensions work
15:18smacleodmars: are you not using trellobug?
15:18marssmacleod: not yet, I haven't set it up
15:19marsI will next time I hack on some cards. Doing pair programming right now
15:20smacleodooo, with?
15:48marsdavidwalsh: hgext.hgmo.mozbuildinfowebcommand:243
16:06marsimadueme: smacleod, zalun:
16:25imaduemeckolos: just found another problem, which I emailed you about. Sorry to bust the bubble :)
16:29mcotemars imadueme davidwalsh smacleod: how are you finding Python 3?
16:29mcotejust writing a blog post in which I've touched on all the new stuff we're dealing with in Conduit
16:29smacleodmcote: it's the future. It's better - I like it
16:29mcoteanything specific? :)
16:30mcoteso far I have just mentioned that we're insulated from the 2020 Python 2 deadline
16:30mcoteanything more about what makes it more fun to use would be cool :)
16:30smacleodmcote: unicode is more sensical
16:30smacleodasync await is awesome for async programming
16:30smacleodalthough I haven't used much of asyncio with it, mostly just the tornado stuff
16:31smacleodso yay, coroutines!
16:31mcoteheh I then segue into "tornado was maybe not the best choice at this time" so I may not mention the async part of python 3;)
16:31smacleodwell, tornado is awesome, it being the wrong choice is complex
16:32mcotethe reason behind it being the wrong choice is complex, or the fact that it is overly complex why it's the wrong reason?
16:32* mcote presumes the latter
16:32mcoteI've summed it up as "premature optimization"
16:32mcotealong with react
16:32mcotewhile fully recognizing they are good choices in the right context
16:38marsmcote: +1 for python 3.5. I didn't have a problem with tornado? I think it would support everything we want to do for a microservice, except the swagger stuff.
16:39mcotewell I recall it causing some confusion because of the asynchronous nature
16:39davidwalshimadueme: I just thought of something now that we're no longer proxying; our `fetch()` calls will need a few config updates: { mode: 'cors', credentials: 'same-origin' }
16:39mcotepicking it instead of, say, Flask, almost certainly led to some extra delays in getting stuff working
16:39mcotewithout much of a benefit at this time
16:40smacleodI don't know if it led to delays already
16:40smacleodbut it might lead to them in the future due to unfamiliarity, and less support for other common python libraries which are synchronous
16:41mcotewell again, I recall people saying they had to learn how to write unit tests in an asynchronous way
16:41smacleodAh, ya I guess there was *that* short delay
16:41marsI think it did lead to delays, because learning anything new causes a delay.
16:41smacleodreally, we should have just used elm
16:41mcoteand we shouldn't be learning things just for the sake of learning them right now :)
16:41marsmcote: yep - I had to spend time across 2? sprints to get unit tests working
16:42mars^ that does tornado + OpenAPI?
16:42smacleodmars: that's a client though, not a server
16:43smacleodyou can use tornado's http client with it, to make calls *TO* a swagger api, from tornado, not create a swagger api server with tornado
16:44marslots of tools, interestring:
16:46smacleodmars: ya, I'm not convinced they're all high quality though... I've been looking through most of them
16:47smacleodAlso, I was sad there was no rust...
16:51marsooo, some clojure support though
16:51marsdavidwalsh: back online, lmk when you want to start hacking again
16:53davidwalshmars: My Vidyo room?
16:55smacleodmars / mcote: is scrum-master certification a thing, that mars actually has?
16:55marsdavidwalsh: sure
16:55marssmacleod: yep!
16:56smacleodah, interesting... I didn't realize that
16:56smacleodIt's good that one of us know's what they're doing :)
16:56smacleodknows too
16:58mcoteI have half a mind to try the CSPO training
16:58marssmacleod: I've also been learning and practicing agile techniques for over a decade now
16:59* smacleod wonders how many times mars has cringed while we butchered scrum
16:59mcoteI'm sure there is no expectation from a CSM that a new team will grasp scrum immediately :)
16:59mcotegrasp and/or adapt to
17:01mcotesmacleod: I totally mansplained Scrum to Melissa. Turns out she took Scrum training when at Autodesk :)
17:02smacleodheh. We did something like what we're now doing when I was on the firefox frontend team... it was nowhere near as nice though
17:09imaduemesmacleod: I have that checklist when ur ready
17:39smacleodmars / davidwalsh: something to keep in mind, "invoke format" needs to be the final format before pushing for review / landing still, and it does things like checks python2 vs python3 differently
18:38imaduemeCase instead of Topic?
19:59marssmacleod: could you please r+ and land ?
20:08mcoteI can do that
20:10pulsebotCheck-in: - David Walsh - Conduit: Create Docker container for Mercurial Server (Bug 1346949). r=mars,mcote
20:10pulsebotCheck-in: - David Walsh - Conduit: Create simple mercurial server extension which returns JSON from custom endpoint (Bug 1346949). r=mars,mcote
20:27marsmcote: thanks!
20:28davidwalshI heard talk of me taking over dkl's card
20:28marssmacleod: review for the swagger API is done, ping me if you want to go over it
20:28marsdavidwalsh: It got bumped up to 8 points :/ But we can shave a point or two off of it, if you want.
20:28davidwalshDo we have BMO working here?
20:29mcotepart of it is creating a mountebank
20:29mcotefor BMO
20:29mcotewe may include BMO as a docker container mainly for demoing purposese
20:29mcotethere's also a BMO vagrant setup
20:30mcoteanyway it's not necessarily about you taking over that card, rather that it's almost certainly going to be a miss without help (and probably even with help) given that dkl has been out sick for a couple days
20:30marsdavidwalsh: maybe we can start with copying the pytest stuff from autoland into commitindex?
20:32marsnot sure how much of the dev-* infrastructure code smacleod has put in
20:36marsdavidwalsh: looks like a bunch of stuff has to be added to match autoland. I can go over the list quickly in a few minutes if you want
21:16mcoteI'm not crazy about "case"
21:16mcotesmacleod doesn't like "topic"
21:17mcoteno one seems to like "proposition" ;)
21:17mcote"solution" seems overused
21:17mcote"motion"? :)
21:25mcoteI think gerrit's equivalent is a "change"
21:26mcoteand their version of an iteration is a "patch set"
21:27mcote"proposal" is another possibility
21:28mcoteheh "grist"
21:28imaduemeI like Proposition haha
21:29imaduemeI think it will positively impact Mozilla's culture and help us discuss code more professionally :)
21:29mcoteheh not sure what percentage of people will immediately think "logical proposition" when they see/hear it though :)
21:30mcote"concern" heh
21:30imaduemeHey steven, can you review my Proposition?
21:30mcoteI have a new iteration of my proposition for you
21:31mcotethis time it's much clearer
21:35imaduemeHonestly can we just call them MozRequests?
21:35imaduemeThat seems to fit
21:36imaduemeOr even just pull requests
21:45mcoteheh no more "moz" :)
21:46mcoteplus as we discussed today they aren't necessarily requests, at least not yet
21:47mcoteof course there's also "set" or "series"
21:47mcoteas now I'm having trouble remembering why "series" was bad... maybe just because we end up with a series of series?
21:48imaduemeCan we call them Bundles?
21:49imaduemeNo one's doing ruby here anyway :)
15 Mar 2017
No messages
Last message: 11 days and 13 hours ago