mozilla :: #conduit

19 May 2017
14:53mcotealmost like imadueme just said that ;)
14:53* imadueme
14:53mcotetoo quiet :)
14:56imaduemeah is my mic low again?
14:57davidwalshimadueme: it usually is
15:11marsimadueme: davidwalsh, I'm thinking we should switch the lando-ui project URL in local development enviroments to .test, as in http://lando.test:7777 or just http://lando.test.
15:12marsimadueme: davidwalsh, however I can see that we have some code in flight. What should land first: the stuff in flight, or the switch to .test?
15:45imaduememars: go for it :) (although I think we must use port 7777 or at least not port 80)
15:54marsimadueme: because URLs matter, does lando-ui require hacking /etc/hosts?
15:54marsdavidwalsh: ^?
15:55imaduememars: Yes - whatever you set the session cookie domain environment variable to must be the same hostname:port you use to access the server in development.
15:56marsimadueme: then should we be using the same tinyproxy setup as the demo environment?
15:56marsto save people from having to hack /etc/hosts
15:58imaduememars: I was just about to say that I don't think it is worthwhile to add that, since there are cases where we will want to test in different browsers and such and in this case there is only 1 container that runs standalone. It makes more sense for the demo imo where we pull many different containers to do a full integration test
15:58imadueme(and it would break my windows browser --> vm server setup )
15:59imadueme(that's the real reason haha, for the demo I have to use the vm browser :P)
16:26imaduemereviewed mars :)
17:09imaduemedavidwalsh: I imagine smacleod would want us to use the package he made here:,
17:09imadueme *
17:31marsdkl: is the bmo container configured to use "" as the URL? I would like to change it to "bmo.test" in the demo environment, but renaming the service in docker-compose.yml didn't work.
17:34dklmars: yes the bmo config has hard coded
17:34dklwe can just rebuild mozilla-conduit/docker-bm
17:35dklshould i go ahead and do that?
17:35marsdkl: I'd like to know what the team thinks of the PR to rename the services, first
17:35marsdkl: I'll push it
17:38marsdkl: smacleod had trouble with hostnames in the past, so I'd like to wait until next week to get his opinion, too
17:38* imadueme what exactly does a PHID stand for?
17:39marsno idea, said the same thing yesterday. Reading the API docs? :)
17:39imaduemeyup, it sounds like just a general identifier for anything in phab :P davidwalsh?
17:39marsdkl might know
17:40* mars stepping out for an appointment
17:43dklimadueme: as he said. everything in phab is an object more or less. and each thing has a phid
17:43dklthe phid also has the object type in the string
17:43dklso a user would be PHID-USER-djfksjdfklsjffjdfjsd
17:44dklprojects are PHID-PROJ-blah
17:44imaduemeso the PHID for a revision is DREV-1234456 but the actual "id" is just 1234456?
17:44dklyou can use the /api/phid.lookup endpoint to find things generally
17:44dkltheir are PHIDs and there are internal ids which are simpler
17:45dklso a revision can be D1 (internal) and also is PHID-DREV-1234556
17:46imaduemeHmm, so dkl If I have a revision "D17957" is there a formal way of converting that to either an ID or PHID? (Is striping the letters the "right" way to get the id?).
18:01dklimadueme: nope you need to use the API to lookup the real PHID. The PHID is a PHID-<TYPE>-<random string>
18:33davidwalshdkl: I&#39;m just seeing your PR now
18:33davidwalshDidn&#39;t get an email or anything, I don&#39;t think
18:39dkldavidwalsh: the PR that i merged already you mean?
18:40davidwalshdkl: No, this
18:40dklah that yes
18:40dklum. i have does some testing locally using the new endpoints and seems to work. I suppose it would be nice to get a second set of eyes since it was written somewhere else.
18:41dkli didnt see any glaring issues myself such as security holes, etc.
20:49imaduemeCurious about everyone&#39;s thoughts on this situation. Not super important but I guess we&#39;re getting to the point where details matter:
20:51imaduemeLet me know your thoughts?
20:53marsimadueme: let&#39;s go with option 3 for now. We can see if option 2 becomes possible after a security review (do we need that level of security?), and/or if another feature happens to make the development of a 403 code easier.
20:55dklvote: option 3
20:57imaduemeSounds good to me! Safe and simple to implement :) Just wanted a second opinion
21:02dklas a general rule you do not want to give out more information than is necessary.
21:09* imadueme will do
20 May 2017
No messages
Last message: 91 days and 5 hours ago