Options:
- # Session Start: Thu Oct 23 00:00:00 2014
- # Session Ident: #testing
- # [00:00] <jgraham> I have literally no idea what webdriver is trying to specify about paths
- # [00:00] <AutomatedTester> paths for what?
- # [00:00] <jgraham> Is it intended that /foo/bar(path in spec) is valid?
- # [00:00] <jgraham> For commands
- # [00:01] <jgraham> The testsuite has /wd/hub as a prefix hardcoded
- # [00:01] <jgraham> But I can't even tell if that's supposed to be allowed
- # [00:01] <AutomatedTester> thats a legacy thing from the OSS project
- # [00:01] <AutomatedTester> iirc that was an action for Simon to figure out and document
- # [00:02] <AutomatedTester> but he has been MIA
- # [00:02] <AutomatedTester> well... till just before TPAC
- # [00:02] <jgraham> So what does "The URLs given in this specification MUST have the same path prefix. Implementations are free to redirect those as necessary." mean?
- # [00:02] <AutomatedTester> good question
- # [00:02] * AutomatedTester tries to decipher what Simon put
- # [00:03] <jgraham> (I honestly have no idea what the intent of that sentence is)
- # [00:03] <AutomatedTester> it looks like its a padding sentence and adding no value
- # [00:03] <AutomatedTester> but maybe that is me
- # [00:03] * Quits: bhill2 (~bhill2@public.cloak) (Client closed connection)
- # [00:04] <jgraham> It has an RFC2119 keyword in it!
- # [00:04] <AutomatedTester> jgraham: go have some chocolate and let me see what he meant :)
- # [00:05] <jgraham> FWIW without having actual well-known urls it seems like you always need configuration to connect to a remote
- # [00:05] <jgraham> So I don't know why you'd want that
- # [00:05] <AutomatedTester> agreed
- # [00:06] * Joins: bhill2 (~bhill2@public.cloak)
- # [00:06] * Quits: Lachy__ (~Lachy@public.cloak) ("My MacBook Pro has gone to sleep. ZZZzzz…")
- # [00:07] <jgraham> (the reaon I noticed is that the testsuite seems to hardcode this)
- # [00:07] <AutomatedTester> jgraham: I am going raise a bug about this, it seems silly
- # [00:07] * Quits: bhill2 (~bhill2@public.cloak) (Client closed connection)
- # [00:07] <jgraham> Thanks
- # [00:07] <jgraham> There are a lack of micro usb cables in my house
- # [00:08] <AutomatedTester> because it reads that IE could do /webdriver/foo/bar and Firefox could have /marionette/foo/bar
- # [00:08] <AutomatedTester> and that either is ok...
- # [00:14] * Joins: bhill2 (~bhill2@public.cloak)
- # [00:14] * Quits: bhill2 (~bhill2@public.cloak) (Client closed connection)
- # [00:15] * Joins: bhill2 (~bhill2@public.cloak)
- # [00:15] * Quits: bhill2 (~bhill2@public.cloak) (Client closed connection)
- # [00:17] * Joins: bhill2 (~bhill2@public.cloak)
- # [00:22] <ato> Perhaps we need to differentiate a bit here.
- # [00:22] <AutomatedTester> ato: jgraham https://www.w3.org/Bugs/Public/show_bug.cgi?id=27136
- # [00:23] <ato> It's a genuinely useful thing to have a WebDriver intermediary remote (proxy) that listens on some specific subpath on a host
- # [00:23] <ato> But the situation right now is that the final remotes (the drivers) all use different base paths and that client bindings must hard code one for each.
- # [00:26] <jgraham> What's the use case for the proxy with multiple paths?
- # [00:28] <ato> Say you have an existing httpd and want to run a WebDriver proxy on something else than /session.
- # [00:28] <ato> (serve.py with a WebDriver handler to avoid CORS is an example of this.)
- # [00:29] * Quits: mdas (~mdas@public.cloak) ("Leaving...")
- # [00:32] <jgraham> Why would serve.py want to actually handle webdriver commands?
- # [00:33] <jgraham> AutomatedTester: You couldn't wire up some in-sec bug reporting tool, could you?
- # [00:33] <jgraham> "Removing a session" doesn't specify what the response should be
- # [00:34] * Quits: Ms2ger (~Ms2ger@public.cloak) ("nn")
- # [00:34] <AutomatedTester> jgraham: I have never found one, the DOM ones seems to be tied to their markup tool
- # [00:34] <AutomatedTester> jgraham: recommend one and I will wire it up
- # [00:37] <ato> jgraham: That's a little besides the point, but in this case to avoid the restrictions of CORS. If you want to make a call to a server on a different port (or even different host) that will act as a proxy, and it's not given that you have the opportunity to set use the same base path as the server you're proxying to for that.
- # [00:38] <jgraham> I'm struggling to see a real life situation in which you want to run a webdriver proxy and other services on the same origin
- # [00:39] <jgraham> Apart from anything else it gives js on those other services webdriver access, which seems like a pretty bad idea
- # [00:40] <jgraham> (I guess you could set up http auth, but still)
- # [00:40] <ato> I think it's a pretty good idea for some things, because it gives your in-document JS control over the OOP browser session.
- # [00:41] <ato> If you want to write a test for window resize evens it's a useful thing to be able to both resize the window and make the assertion in the same context.
- # [00:41] <ato> events*
- # [00:42] <jgraham> Yeah, so I suppose for the specific use case of wpt it could make for an easier setup
- # [00:42] <ato> I'd like to see the WebDriver servers use the same base path by default however.
- # [00:43] <ato> If individual servers choose to provide an option to override that I guess that's fine, and not something it's necessary to put in a spec.
- # [00:43] <ato> Like chromedriver's --base-path=/foo flag.
- # [00:44] <ato> Which will make the current URL accessible as GET /foo/{session_id}/url.
- # [00:48] <jgraham> Wait, what?
- # [00:48] <jgraham> /foo/session/{} seems OK
- # [00:48] <jgraham> replacing /session doesn't
- # [00:53] <ato> Why is /session a requirement in your eyes?
- # [00:54] <jgraham> Because "GET /foo" to create a session and "DELETE /foo/{id}" to delete one don't make any sense
- # [00:55] <jgraham> If making sense there isn't a requirement I think /session should be dropped entirely
- # [00:58] <ato> You suggest we mandate /session from the drivers, right? And say that they may give an option to change the base path, which is /, to something else so that the full path to the session would become /foo/session?
- # [01:00] <MikeSmith> ato: thanks, I'll take a look at that PR on critic
- # [01:01] <jgraham> ato: I think that is the least insane option if they have to be allowed to change the base path
- # [01:01] <ato> WFM
- # [01:01] <ato> I was just curious about requiring them to keep /session intact. My solution would basically be to leave it unspecified.
- # [01:03] <ato> But I suppose we see what happens if we don't say anything: chromedriver uses /wd/url, Selenium uses /wd/hub/session, and FirefoxDriver /session.
- # [01:03] <ato> :D
- # [01:04] <jgraham> I think the spec is trying to specify this, but failing to do so
- # [01:04] <jgraham> That's the bug AutomatedTester raised
- # [01:04] <ato> Thanks for that.
- # [01:04] <ato> I remember reading it and being confused.
- # [01:15] * Quits: ryan (~ryan52@public.cloak) (Ping timeout: 180 seconds)
- # [01:31] * Joins: Lachy (~Lachy@public.cloak)
- # [01:33] * Joins: scott_gonzalez (~scott_gonzalez@public.cloak)
- # [01:34] * Quits: plh (plehegar@public.cloak) ("Leaving")
- # [01:46] * Quits: MK (kawamori@public.cloak) (Ping timeout: 180 seconds)
- # [02:06] * Quits: ptressel (~chatzilla@public.cloak) ("bbl")
- # [02:10] * Quits: AutomatedTester (~AutomatedTester@public.cloak) (Client closed connection)
- # [03:06] * Joins: bhill2_ (~bhill2@public.cloak)
- # [03:12] * Quits: bhill2 (~bhill2@public.cloak) (Ping timeout: 180 seconds)
- # [03:13] * Quits: bhill2_ (~bhill2@public.cloak) (Ping timeout: 180 seconds)
- # [03:13] * Quits: Lachy (~Lachy@public.cloak) ("Textual IRC Client: www.textualapp.com")
- # [03:16] * Joins: MK (kawamori@public.cloak)
- # [03:29] * Quits: ArtB (~abarsto@public.cloak) ("Leaving.")
- # [03:40] * Quits: MK (kawamori@public.cloak) (Ping timeout: 180 seconds)
- # [03:42] * Joins: MK (kawamori@public.cloak)
- # [03:57] * Quits: MK (kawamori@public.cloak) (Ping timeout: 180 seconds)
- # [06:15] * Joins: MK (kawamori@public.cloak)
- # [06:18] * Joins: MK_ (kawamori@public.cloak)
- # [06:24] * Quits: MK (kawamori@public.cloak) (Ping timeout: 180 seconds)
- # [06:26] * Joins: MK (kawamori@public.cloak)
- # [06:28] * Joins: MK__ (kawamori@public.cloak)
- # [06:31] * Quits: MK_ (kawamori@public.cloak) (Ping timeout: 180 seconds)
- # [06:34] * Quits: MK (kawamori@public.cloak) (Ping timeout: 180 seconds)
- # [07:21] * Quits: scott_gonzalez (~scott_gonzalez@public.cloak) (Client closed connection)
- # [07:23] * Quits: MK__ (kawamori@public.cloak) ("Leaving...")
- # [07:53] * Joins: ptressel (~chatzilla@public.cloak)
- # [09:45] * Joins: Ms2ger (~Ms2ger@public.cloak)
- # [10:21] <MikeSmith> ato: https://critic.hoppipolla.co.uk/r/2949 may not be perfect but I think it's good enough. Works well for me at least in my environment. I'd be happy if it were landed so that I can use it next week to demo at the testing breakout session at TPAC.
- # [10:36] <MikeSmith> I'm pretty sure the ChromeDriver docs are trolling me
- # [10:37] <MikeSmith> ーan executable downloaded from the Chromium project which acts as a bridge between "chrome" and the "driver". This executable is called "chromedriver", but we'll try and refer to it as the "server" in this page to reduce confusion.
- # [10:38] <MikeSmith> "Simply start up the ChromeDriver executable (that works as a server), create a client, and away you go"
- # [10:39] <MikeSmith> I also like how the section on reporting bugs is titled, "Think you've found a bug?"
- # [10:39] <MikeSmith> I think these docs have a chip on their shoulder
- # [10:40] <Ms2ger> You appear to have found a bug. Would you like help with that?
- # [10:41] <MikeSmith> heh
- # [10:41] <MikeSmith> "We realise that these terms are confusing, so please feel free to suggest something better!"
- # [10:42] <MikeSmith> the sectioning containing that text is mistitled
- # [10:42] <MikeSmith> for consistency I think they meant to title it, "So you think you can come up with less-confusing terms?"
- # [10:43] * MikeSmith files a docs bug
- # [10:44] <Ms2ger> Ha
- # [11:17] <ato> MikeSmith: Sounds good!
- # [11:22] <jgraham> Maybe they should call sometimes call the server "the client"
- # [11:22] <MikeSmith> ato: nice work on that btw, and thanks
- # [11:23] <MikeSmith> jgraham: good brainstorming
- # [11:23] <MikeSmith> so hey I still don't understand how to run firefox from wptrunner
- # [11:24] <MikeSmith> I realize I can run it with mach in my mozilla working directory
- # [11:24] <MikeSmith> but I'd like to demo running it directly from wptrunner of possible
- # [11:30] * Joins: Lachy (~Lachy@public.cloak)
- # [11:36] <jgraham> MikeSmith: wptrunner --binary=~/develop/gecko-dev-1/obj-x86_64-unknown-linux-gnu/dist/bin/firefox --tests=/home/jgraham/develop/web-platform-tests/ --metadata=/home/jgraham/develop/web-platform-tests/ --run-by-dir --log-mach=- --prefs-root=~/develop/gecko-dev/testing/profiles/ --processes=1 --include=dom --log-mach-level=debug
- # [11:36] <jgraham> Is an actual command from my history
- # [11:37] <jgraham> --run-by-dir and --processes aren't needed
- # [11:37] <jgraham> But --prefs-root is
- # [11:37] <jgraham> (also not --include unless you want that)
- # [11:43] <MikeSmith> jgraham: ah ok thanks I didn't have --prefs-root
- # [11:44] <jgraham> Yeah, it kind of sucks, but I don't know what the best thing do do is
- # [11:45] <MikeSmith> hmm I don't know what a press file looks like. is there a default somewhere
- # [11:45] <jgraham> It could try to download a copy, or we could include a copy, or something
- # [11:45] <MikeSmith> Ok
- # [11:45] <jgraham> You mean prefs?
- # [11:45] <MikeSmith> yeah
- # [11:46] <jgraham> You need to use one in your source tree. $SOURCE_ROOT/testing/profiles/
- # [11:46] <MikeSmith> hai
- # [11:53] * Joins: darobin (rberjon@public.cloak)
- # [12:17] * Quits: Ms2ger (~Ms2ger@public.cloak) (Ping timeout: 180 seconds)
- # [12:22] * Quits: Lachy (~Lachy@public.cloak) ("My MacBook Pro has gone to sleep. ZZZzzz…")
- # [12:27] * Joins: abarsto (~abarsto@public.cloak)
- # [12:27] * abarsto is now known as ArtB
- # [12:53] * Joins: Lachy (~Lachy@public.cloak)
- # [13:15] * Joins: Ms2ger (~Ms2ger@public.cloak)
- # [13:21] <jgraham> Bah, this whole partial configuration overriding stuff doesn't work too well with ssl support
- # [13:22] <jgraham> Because different options need totally different keys
- # [13:22] <jgraham> I guess the eaiest thing to do is go super-verbose and include all the possibilities in the default configuration
- # [13:59] * Quits: Lachy (~Lachy@public.cloak) ("My MacBook Pro has gone to sleep. ZZZzzz…")
- # [14:16] * Joins: Lachy (~Lachy@public.cloak)
- # [14:21] <MikeSmith> jgraham: what does the --product switch do?
- # [14:23] <ato> It chooses the correct product? wptrunner can't deduce which browser specification to use from the binary alone.
- # [14:26] * Joins: Lachy_ (~Lachy@public.cloak)
- # [14:28] * Quits: Ms2ger (~Ms2ger@public.cloak) (Ping timeout: 180 seconds)
- # [14:29] <MikeSmith> ato: but it's not required and jgraham didn't have it in the command he pasted in above
- # [14:29] <ato> Hm I thought it was. My knowledge is outdated then.
- # [14:31] * Quits: Lachy (~Lachy@public.cloak) (Ping timeout: 180 seconds)
- # [14:45] * Joins: Lachy (~Lachy@public.cloak)
- # [14:45] * Quits: Lachy_ (~Lachy@public.cloak) (Client closed connection)
- # [14:47] <jgraham> MikeSmith: Firefox is the default
- # [14:55] <ato> And here I've been typing "--product firefox" for months. You owe me back some characters.
- # [14:56] <MikeSmith> heh
- # [15:00] * Joins: plh (plehegar@public.cloak)
- # [15:00] * Joins: plehegar__ (plehegar@public.cloak)
- # [15:01] * Quits: plehegar__ (plehegar@public.cloak) (Client closed connection)
- # [15:15] * Quits: Lachy (~Lachy@public.cloak) ("My MacBook Pro has gone to sleep. ZZZzzz…")
- # [15:19] * Joins: Lachy (~Lachy@public.cloak)
- # [15:20] * Quits: Lachy (~Lachy@public.cloak) ("My MacBook Pro has gone to sleep. ZZZzzz…")
- # [15:21] * Joins: Lachy (~Lachy@public.cloak)
- # [15:27] * Quits: Lachy (~Lachy@public.cloak) ("My MacBook Pro has gone to sleep. ZZZzzz…")
- # [15:30] * Joins: gitbot (~gitbot@public.cloak)
- # [15:30] -gitbot:#testing- [web-platform-tests] jgraham pushed 1 new commit to master: https://github.com/w3c/web-platform-tests/commit/2ea4b62e7e9a27b9d158e21cbe7d155aabbaae2f
- # [15:30] -gitbot:#testing- web-platform-tests/master 2ea4b62 James Graham: Update to latest wptserve
- # [15:30] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
- # [15:39] <MikeSmith> is TBPL format a mozilla thing?
- # [15:39] <MikeSmith> ah tinderbox
- # [15:40] <MikeSmith> Tinderboxpushlog
- # [15:43] <jgraham> In the beginning there was the CI system and that CI system was tinderbox
- # [15:43] <jgraham> And then tinderbox begot tbpl and tbpl begot treeherder
- # [15:45] <jgraham> But the format is still called tbpl format even though tinderbox has been switched off for some time now
- # [15:46] <MikeSmith> jgraham: OK
- # [15:47] <MikeSmith> so the output from the --log-html format is known not to be right?
- # [15:48] <MikeSmith> it just lists all the tests and claims all of them passed even though the log-mach and raw output show that there were failures
- # [15:49] <jgraham> MikeSmith: I think the honest answer is that I don't remember.
- # [15:49] <MikeSmith> ok
- # [15:49] * Joins: Lachy (~Lachy@public.cloak)
- # [15:49] <MikeSmith> I'm just playing around with it anyway
- # [15:50] <jgraham> I think we are going to release a new mozlog today or so with improvements to the HTML format, so retrying after we do that might help
- # [15:50] <MikeSmith> I guess to contribute something useful I could write some PRs with minor updates to the .rst docs
- # [15:50] <MikeSmith> jgraham: k
- # [15:50] <jgraham> Honestly I don't know how useful --log-html is for wpt since there tend to be an awful lot of tests and subtests
- # [15:50] <MikeSmith> right
- # [15:51] <MikeSmith> the --log-mach output is nice anyway
- # [15:51] <MikeSmith> as far as having something readable
- # [15:52] <jgraham> Yeah, the log-tbpl format is specifically designed to work well with the crazy regexp our various log parsers use
- # [15:53] <jgraham> although the long term goal is to make those operate on the json directly
- # [15:53] <MikeSmith> ok
- # [15:54] <jgraham> (we're half way there for wpt; the actual status of the job is set from the JSON, but the summary of any errors in the web UI comes from parsing the --log-tbpl output)
- # [15:55] <MikeSmith> I think it's pretty impressive how far along all this already, as far as having something that works across more than one browser engine
- # [16:00] * Joins: anssik (~uid10742@public.cloak)
- # [16:02] <jgraham> It feels like it could be run against Blink in CI with very little effort
- # [16:03] <jgraham> But it needs someone to actually do that
- # [16:04] * Joins: mdas (~mdas@public.cloak)
- # [16:11] <MikeSmith> jgraham: good thing you'll be at GTAC and have time to talk to Chrome testing people and convince them
- # [16:13] <MikeSmith> do we really not have somebody on the blink team that's already semi-aware of the core wpt work and thinking about how to integrate it into their automation?
- # [16:14] <MikeSmith> a Blink counterpart for Dirk Pranke
- # [16:15] <MikeSmith> btw is FirefoxDriver still an active thing, or has it been completely superseded by Marionette
- # [16:16] <jgraham> ato: ^
- # [16:17] * Joins: Ms2ger (~Ms2ger@public.cloak)
- # [16:25] <Ms2ger> denis, what's the loop variable at https://github.com/w3c/web-platform-tests/blob/servo/html/dom/documents/resource-metadata-management/document-readyState.html#L26 for?
- # [16:26] <jgraham> Apart from that I was going to comment that it was a nice test :)
- # [16:26] <Ms2ger> And https://github.com/w3c/web-platform-tests/blob/master/html/dom/documents/resource-metadata-management/document-readyState.html#L26
- # [16:27] * Ms2ger should test his links
- # [16:30] <denis> it should be removed Ms2ger
- # [16:30] * Joins: Lachy_ (~Lachy@public.cloak)
- # [16:31] <Ms2ger> Knock yourself out ;)
- # [16:31] <Ms2ger> I'll review!
- # [16:32] <denis> ok Ms2ger I'll take care of it in a bit
- # [16:33] <Ms2ger> Thanks!
- # [16:33] <MikeSmith> well this is weird, http://web-platform.test:8000/tools/runner/index.html no long works in my environment
- # [16:34] <MikeSmith> instead http://web-platform.test:8000/tools/runner/tools/runner/index.html does
- # [16:34] <Ms2ger> I believe one jgraham recently touched that
- # [16:34] <MikeSmith> something borked
- # [16:34] * MikeSmith looks at the changelog
- # [16:36] * Quits: Lachy (~Lachy@public.cloak) (Ping timeout: 180 seconds)
- # [16:39] <jgraham> Hmm
- # [16:42] <MikeSmith> Ms2ger: yeah and I reviewed it :(
- # [16:42] <Ms2ger> MikeSmith, :)
- # [16:51] * Joins: gitbot (~gitbot@public.cloak)
- # [16:51] -gitbot:#testing- [web-platform-tests] jgraham created jgraham/ssl_support (+1 new commit): https://github.com/w3c/web-platform-tests/commit/94d456679653
- # [16:51] -gitbot:#testing- web-platform-tests/jgraham/ssl_support 94d4566 James Graham: Support running tests with SSL....
- # [16:51] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
- # [16:52] * Joins: gitbot (~gitbot@public.cloak)
- # [16:52] -gitbot:#testing- [web-platform-tests] jgraham opened pull request #1302: Support running tests with SSL. (master...jgraham/ssl_support) https://github.com/w3c/web-platform-tests/pull/1302
- # [16:52] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
- # [16:53] <jgraham> OK, created a PR for SSL support, but I don't think it's necessarily "done" yet, I just want to encourage people to try it out
- # [16:57] <plh> any specific test you recommend for testing the ssl support?
- # [16:58] <jgraham> The PR didn't actually include any fixes for tests
- # [16:58] <jgraham> iirc there's only one test currently that tries to use https. I can fix that and add it to the branch
- # [16:58] <plh> I think we have more than one. others have been disabled
- # [16:58] * plh looks
- # [16:59] <jgraham> There are some wss ones, but that isn't working yet (just because I didn't get to it)
- # [17:00] * Quits: hober (~ted@public.cloak) (Client closed connection)
- # [17:00] <plh> html/editing/dnd/resources/crossorigin.sub.js has some https
- # [17:00] * Joins: hober (~ted@public.cloak)
- # [17:01] <jgraham> Yeah, that's a manual test though
- # [17:01] <jgraham> Or they are all manual tests
- # [17:01] <plh> yes
- # [17:01] <jgraham> MikeSmith: Dammit
- # [17:02] <plh> so, I try to access https://web-platform.test:8000/, I get "SSL connection error"
- # [17:02] <plh> not sure if it's intended or not
- # [17:02] <jgraham> plh: Yeah, the ssl server is typically 8443
- # [17:03] <plh> https://web-platform.test:8443/ is untrusted
- # [17:03] <plh> I guess we'll have to tell folks to trust it anyway
- # [17:04] * Quits: Lachy_ (~Lachy@public.cloak) (Lachy_)
- # [17:06] * Joins: Lachy (~Lachy@public.cloak)
- # [17:08] <plh> so I can report that it works on chrome and firefox, once you get passed the untrusted/warning boxes
- # [17:13] * Quits: plh (plehegar@public.cloak) ("Leaving")
- # [17:19] * Joins: simons (~simons@public.cloak)
- # [17:20] <simons> Is there a good guide for working with mercurial that anyone here can recommend?
- # [17:20] <simons> Because I have a pile of edits to make to the webdriver spec, and have spent the day being frustrated by hg
- # [17:21] <jgraham> simons: Put the whole thing in git instead?
- # [17:21] <jgraham> (I'm serious, we can do that)
- # [17:22] <jgraham> Dunno where AutomatedTester is
- # [17:22] <simons> I feel I should probably be quad-lingual with source control systems
- # [17:22] <simons> Just knowing a decent hg workflow would be Very Useful
- # [17:23] <jgraham> At some point in the past http://hginit.com/ was supposed to be good
- # [17:23] <ato> MikeSmith: FirefoxDriver is the official WebDriver implementation for Firefox. It's being superseded by Marionette, but we're not there yet.
- # [17:23] <jgraham> I don't know if it contains modern features though
- # [17:24] <jgraham> The first rule is "don't listen to anyone at Mozilla who tries to teach you their hg workflow"
- # [17:24] <simons> If it covers bookmarks I'll be fine
- # [17:24] <simons> I believe the same applies at facebook
- # [17:24] <ato> MikeSmith: As it so happens wptrunner uses Marionette to drive Firefox because it's the future.
- # [17:24] <jgraham> Because there's about a 99% chance they'll say "mq" at some point
- # [17:24] <simons> Helpful
- # [17:24] <simons> I've always wanted a message queue in the middle of my source control process
- # [17:24] <ato> Yes to put the whole thing in git.
- # [17:25] <ato> We don't have time to frustrate about hg.
- # [17:25] <simons> Meh. Once I understand the system, I'm sure I'll be equally grumbly about hg as I am about git
- # [17:26] <jgraham> http://gregoryszorc.com/blog/2014/06/23/please-stop-using-mq/ might be useful, although like all things gps writes, it's about 25% anti-git ranting
- # [17:26] <ato> I'm cleverly avoiding hg at Mozilla by using the git mirror.
- # [17:28] <Ms2ger> Also don't ask jgraham, because it'll be 100% anti-hg ranting
- # [17:28] <simons> So far, my workflow is: hg clone blah blah; cd blah; hg bookmark -r default magic; hg up magic; touch ttt; hg add ttt; hg commit -m 'Look'
- # [17:28] <simons> And then it falls apart
- # [17:29] <simons> I've got access to hg committers. They just happen to be asleep. So the access is theoretical :)
- # [17:29] <simons> I'd been under the impression that an hg bookmark was exactly analogous to a git branch
- # [17:29] <simons> This appears to be fallacious
- # [17:30] <Ms2ger> simons, assuming exact equivalence is always going to be a fallacy
- # [17:30] <simons> Well, I know that now
- # [17:30] <Ms2ger> If you have access to the upstream repo, hg clone blah blah; cd blah; touch ttt; hg add ttt; hg commit -m 'Look'; hg push will just work
- # [17:31] * Joins: bhill2 (~bhill2@public.cloak)
- # [17:31] <simons> I really wanted a handful of diffs in flight at the same time
- # [17:31] <simons> In git, I'd keep "master" pristine
- # [17:31] <simons> and just work on a series of branches
- # [17:32] <simons> Which I'd rebase on master and merge back in when I was ready to push (to preserve a linear history)
- # [17:32] <jgraham> simons: Yeah, so that workflow is possible with bookmarks
- # [17:32] <simons> Excellent. Can I treat "default" as a bookmark?
- # [17:32] <jgraham> But there are major missing features that make it harder than it needs to be
- # [17:32] <jgraham> Yeah, default is a bookmark
- # [17:32] <jgraham> But there isn't anything that corresponds to origin/master
- # [17:32] <simons> How come it doesn't show up in "hg bookmarks"?
- # [17:33] <jgraham> I thought it did…
- # [17:33] <jgraham> I dont' do this much
- # [17:33] <simons> Nope
- # [17:33] <simons> :)
- # [17:33] <simons> Neither do I. It's been an experience
- # [17:34] <jgraham> Ms2ger: I'm not sure that I've made any anti-hg rants here
- # [17:34] <jgraham> Although I think simons should just switch the repo to git :)
- # [17:34] <Ms2ger> I didn't say it was here :)
- # [17:35] * Joins: gitbot (~gitbot@public.cloak)
- # [17:35] -gitbot:#testing- [web-platform-tests] deniak created denis/clean-readystate-test (+1 new commit): https://github.com/w3c/web-platform-tests/commit/1db739a055cc
- # [17:35] -gitbot:#testing- web-platform-tests/denis/clean-readystate-test 1db739a Denis Ah-Kang: remove unused var
- # [17:35] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
- # [17:37] * Joins: gitbot (~gitbot@public.cloak)
- # [17:37] -gitbot:#testing- [web-platform-tests] deniak opened pull request #1303: remove unused var (master...denis/clean-readystate-test) https://github.com/w3c/web-platform-tests/pull/1303
- # [17:37] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
- # [17:38] <ato> Should /testharness_runner.html in wpt be in .gitignore?
- # [17:39] <simons> The most amusing lesson I've learnt today: there are more hg workflows than git ones. And there are as many git workflows as there are language choices on the Downfall DVD.
- # [17:39] <simons> A truly bewildering number
- # [17:42] * Joins: gitbot (~gitbot@public.cloak)
- # [17:42] -gitbot:#testing- [web-platform-tests] jgraham closed pull request #1303: remove unused var (master...denis/clean-readystate-test) https://github.com/w3c/web-platform-tests/pull/1303
- # [17:42] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
- # [17:51] * Joins: gitbot (~gitbot@public.cloak)
- # [17:51] -gitbot:#testing- [web-platform-tests] deniak deleted denis/clean-readystate-test at 1db739a: https://github.com/w3c/web-platform-tests/commit/1db739a
- # [17:51] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
- # [17:51] * Quits: bhill2 (~bhill2@public.cloak) (Client closed connection)
- # [17:56] * Joins: bhill2 (~bhill2@public.cloak)
- # [18:03] <simons> Being schooled on hg by one of the mercurial people at work. It's a revelation
- # [18:03] <simons> We don't need to move the repo :)
- # [18:03] * Joins: AutomatedTester (~AutomatedTester@public.cloak)
- # [18:04] * jgraham looks forward to the gospel according to simons
- # [18:04] * Joins: plh (plehegar@public.cloak)
- # [18:06] <simons> It's being made to sound very simple
- # [18:07] <jgraham> haha
- # [18:07] <simons> I think I understand what I'm doing now. The key, it turns out, appears to be creating multiple heads in the same branch
- # [18:08] <simons> And understanding that a bookmark is a pointer to a revision, rather than being an actual branch
- # [18:08] <jgraham> Yeah, a bookmark is like a git branch
- # [18:08] <simons> No
- # [18:08] <simons> That's what got me into this mess :)
- # [18:09] <jgraham> A git branch is a pointer to a revision…
- # [18:09] <simons> A git branch is more equivalent to an hg head
- # [18:09] <simons> But changes in one branch don't effect other branches
- # [18:09] <simons> Whereas, since a bookmark is a pointer to a revision, you can royally screw things up in hg
- # [18:10] <jgraham> Now when you say "branch" I have no idea what you are talking about :)
- # [18:10] <simons> :)
- # [18:10] <simons> git branches are independent of one another
- # [18:10] <jgraham> Right
- # [18:10] <jgraham> hg bookmarks aren't?
- # [18:11] <simons> Not always :)
- # [18:11] <jgraham> WTF?
- # [18:11] <simons> They are if the diffs they point to are on different heads
- # [18:12] <jgraham> So if you have
- # [18:12] <jgraham> A - B
- # [18:12] <jgraham> \_ C
- # [18:12] <jgraham> and have two bookmarks pointing to B
- # [18:12] <jgraham> and then add B-D
- # [18:13] <jgraham> Moving one bookmark will move both?
- # [18:13] <simons> Where are your bookmarks?
- # [18:13] <jgraham> (but not if one points at B and one at C ofc)
- # [18:14] <jgraham> Well let's say that A has "a", B has "b" and "b'" and C has "c"
- # [18:14] <jgraham> and you add a new commit to B called D
- # [18:14] <simons> A commit will move the current bookmark
- # [18:15] <jgraham> So typically b or b' in that case?
- # [18:15] <jgraham> But not both?
- # [18:15] <simons> So, if you're on B and commit, the "b" bookmark will now point to "D"
- # [18:15] <jgraham> Right
- # [18:15] <jgraham> So that sounds like git branches
- # [18:16] <jgraham> What shows the difference?
- # [18:16] <simons> But if you do something in "a", I _think_ it ends up being reflected in both "b" and "c"
- # [18:16] <simons> I'm having a play right now
- # [18:16] <jgraham> Wait, what?
- # [18:17] <jgraham> What could you do to A that would possibly move other bookmarks?
- # [18:17] <simons> (like I said, I'm about to have aplay)
- # [18:17] <simons> because "a" isn't a head.
- # [18:17] <simons> (I think)
- # [18:17] <simons> I'm a kinesthetic learner.
- # [18:18] <simons> Will have this cracked in about 10 minutes now
- # [18:19] * jgraham notes that the theory about different people having different types of learning has been largely discredited
- # [18:19] * Quits: Lachy (~Lachy@public.cloak) ("My MacBook Pro has gone to sleep. ZZZzzz…")
- # [18:19] <jgraham> (or at least I think I saw that somewhere)
- # [18:19] <simons> I definitely learn better by doing
- # [18:20] <jgraham> Sure, but I think the idea is that applies more or less to everyone
- # [18:22] <Ms2ger> simons, try teaching
- # [18:22] <simons> Ms2ger: ?
- # [18:22] <Ms2ger> I hear that works even better :)
- # [18:22] <simons> Oh yeah. That forces you to actually know things
- # [18:22] * Joins: gitbot (~gitbot@public.cloak)
- # [18:22] -gitbot:#testing- [web-platform-tests] mbrubeck closed pull request #1300: Update TA11.1-11.2 to use All_Pointer_Events pattern and eliminate timer (master...submission/Microsoft/PointerEvents-TA11Updates) https://github.com/w3c/web-platform-tests/pull/1300
- # [18:22] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
- # [18:22] * Joins: gitbot (~gitbot@public.cloak)
- # [18:22] -gitbot:#testing- [web-platform-tests] mbrubeck pushed 3 new commits to master: https://github.com/w3c/web-platform-tests/compare/b918656cc7c3...ec35bb404e14
- # [18:22] -gitbot:#testing- web-platform-tests/master 88f015d Evgeny Agafonchikov: Fixing All_Pointer_Events issue...
- # [18:22] -gitbot:#testing- web-platform-tests/master 4d498be Jacob Rossi: Fix code style to clarify code flow through conditional
- # [18:22] -gitbot:#testing- web-platform-tests/master ec35bb4 Matt Brubeck: Merge pull request #1300 from InternetExplorer/submission/Microsoft/PointerEvents-TA11Updates...
- # [18:22] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
- # [18:23] <simons> Right. jgraham I was talking nonsense
- # [18:23] <simons> Modifying A won't modify B or C
- # [18:23] <jgraham> OK
- # [18:23] <simons> Though you can rebase for fun and profit
- # [18:24] <jgraham> If you are rebasing for fun, you probably need a hobby. Or help.
- # [18:24] <Ms2ger> Or both
- # [18:24] * Quits: darobin (rberjon@public.cloak) (Client closed connection)
- # [18:28] * Quits: simons (~simons@public.cloak) ("My MacBook Pro has gone to sleep. ZZZzzz…")
- # [18:32] * Joins: Lachy (~Lachy@public.cloak)
- # [18:32] * Quits: Lachy (~Lachy@public.cloak) ("My MacBook Pro has gone to sleep. ZZZzzz…")
- # [18:55] * Joins: bhill2__ (~bhill2@public.cloak)
- # [18:56] * Quits: bhill2 (~bhill2@public.cloak) (Ping timeout: 180 seconds)
- # [19:14] * Quits: anssik (~uid10742@public.cloak) ("Connection closed for inactivity")
- # [19:16] * Quits: bhill2__ (~bhill2@public.cloak) (Client closed connection)
- # [19:18] * Joins: bhill2 (~bhill2@public.cloak)
- # [19:19] * Quits: bhill2 (~bhill2@public.cloak) (Client closed connection)
- # [19:25] * Quits: ArtB (~abarsto@public.cloak) ("Leaving.")
- # [19:27] * Quits: AutomatedTester (~AutomatedTester@public.cloak) (Client closed connection)
- # [19:40] * Joins: bhill2 (~bhill2@public.cloak)
- # [19:40] * Quits: bhill2 (~bhill2@public.cloak) (Client closed connection)
- # [19:40] * Joins: bhill2 (~bhill2@public.cloak)
- # [19:40] * Quits: bhill2 (~bhill2@public.cloak) (Client closed connection)
- # [19:40] * Joins: bhill2 (~bhill2@public.cloak)
- # [19:45] <jgraham> MikeSmith: So I might have said some time ago that I have a fix for your problem
- # [19:46] <jgraham> Turns out that testing it involves rebuilding Firefox
- # [19:46] <jgraham> Which on this machine takes ~forever
- # [19:57] * Joins: abarsto (~abarsto@public.cloak)
- # [19:57] * abarsto is now known as ArtB
- # [20:00] * Quits: bhill2 (~bhill2@public.cloak) (Ping timeout: 180 seconds)
- # [20:14] * Quits: mdas (~mdas@public.cloak) (Client closed connection)
- # [20:15] * Quits: ato (~sid16069@public.cloak) (Client closed connection)
- # [20:15] * Joins: mdas (~mdas@public.cloak)
- # [20:15] * Quits: mkwst___ (~sid395@public.cloak) (Client closed connection)
- # [20:15] * Quits: bterlson______ (~sid23757@public.cloak) (Client closed connection)
- # [20:15] * Quits: scheib (~sid4467@public.cloak) (Client closed connection)
- # [20:15] * Quits: astearns_ (~sid15080@public.cloak) (Client closed connection)
- # [20:15] * Quits: timeless (~sid4015@public.cloak) (Client closed connection)
- # [20:15] * Quits: tobie (~sid5692@public.cloak) (Client closed connection)
- # [20:15] * Quits: bret (~sid12421@public.cloak) (Client closed connection)
- # [20:15] * Quits: krit (~sid15081@public.cloak) (Client closed connection)
- # [20:16] * Quits: sangwhan (~sid12645@public.cloak) (Client closed connection)
- # [20:16] * Joins: bret (~sid12421@public.cloak)
- # [20:16] * Joins: krit (~sid15081@public.cloak)
- # [20:16] * Joins: mkwst___ (~sid395@public.cloak)
- # [20:16] * Quits: tripu (antonio@public.cloak) (Ping timeout: 180 seconds)
- # [20:16] * Joins: sangwhan_ (~sid12645@public.cloak)
- # [20:16] * Joins: scheib (~sid4467@public.cloak)
- # [20:16] * Joins: timeless (~sid4015@public.cloak)
- # [20:16] * Joins: ato (~sid16069@public.cloak)
- # [20:16] * Joins: lmclister______ (~sid13822@public.cloak)
- # [20:16] * Joins: bterlson_______ (~sid23757@public.cloak)
- # [20:16] * Joins: astearns_ (~sid15080@public.cloak)
- # [20:16] * Joins: tobie (~sid5692@public.cloak)
- # [20:18] * Joins: tripu (antonio@public.cloak)
- # [20:18] * Quits: lmclister_____ (~sid13822@public.cloak) (Ping timeout: 180 seconds)
- # [20:18] * lmclister______ is now known as lmclister_____
- # [20:36] * Joins: AutomatedTester (~AutomatedTester@public.cloak)
- # [21:46] * Joins: bhill2 (~bhill2@public.cloak)
- # [21:49] * Joins: Lachy (~Lachy@public.cloak)
- # [22:00] * Joins: bhill2_ (~bhill2@public.cloak)
- # [22:05] * Quits: bhill2 (~bhill2@public.cloak) (Ping timeout: 180 seconds)
- # [22:33] * Joins: simons (~simons@public.cloak)
- # [22:35] * Quits: simons (~simons@public.cloak) ("My MacBook Pro has gone to sleep. ZZZzzz…")
- # [22:46] * Quits: Lachy (~Lachy@public.cloak) ("My MacBook Pro has gone to sleep. ZZZzzz…")
- # [22:50] * Joins: Lachy (~Lachy@public.cloak)
- # [23:00] * Quits: hober (~ted@public.cloak) (Client closed connection)
- # [23:00] * Joins: hober (~ted@public.cloak)
- # [23:17] * Joins: shepazu (schepers@public.cloak)
- # [23:21] * Quits: bhill2_ (~bhill2@public.cloak) (Client closed connection)
- # [23:21] * Joins: bhill2 (~bhill2@public.cloak)
- # [23:22] * Quits: bhill2 (~bhill2@public.cloak) (Client closed connection)
- # [23:24] * Joins: bhill2 (~bhill2@public.cloak)
- # Session Close: Fri Oct 24 00:00:00 2014
The end :)