w3 :: #webdriver

15 Mar 2017
13:17jimevans1can anyone remind me why we use a character array in the payload for the setAlertText command?
13:18AutomatedTesterjimevans1: to align with send_keys
13:19jimevans1but it's no longer aligned.
13:20jgrahamCharacter arrays are a bad idea
13:21jimevans1cf. step 8 of the remote end steps for Element Send Keys: "if text is not a string, return error with error code invalid argument"
13:22jimevansjgraham: no argument from me.
13:23jimevansbut if the reason was to align with element send keys, they no longer align.
13:23jgrahamYEah, I think they should both take strings. File a bug.
13:23jimevanscopy that.
13:29jimevanshttps://github.com/w3c/webdriver/issues/862
13:30jimevansgeckodriver currently expects the array, yes?
13:30jimevans(for the alert case)
13:30* jimevans is too lazy to try it
13:35jgrahamjimevans: It uses the same struct for both cases and it&#39;s SendKeysParameters { pub value: Vec<char> } right now
13:36jgrahamSo that will need to change
13:36jimevansugh. what a nightmare.
13:36jimevanswell that would explain some failures of my webdriver tests against geckodriver...
13:36jimevans:)
13:37jgrahamWell it&#39;s easy to change, I can patch it now :)
13:37jimevanshah!
13:37jimevansas it turns out, i&#39;m sending both `text` as a string and `value` as an array in the payload.
13:38jimevansso that works fine for sendkeys
13:43jgrahamjimevans: https://github.com/mozilla/webdriver-rust/pull/80
13:47AutomatedTesterjgraham: thats a breaking change :/
13:47AutomatedTesterwe will need another coordinated release
13:48AutomatedTesteractually...
13:50jgrahamAutomatedTester: Yeah, it is a breaking change, but c.f. jimevans for how to avoid needining a coordinated release
13:50jgraham13:37 < jimevans> as it turns out, i&#39;m sending both `text` as a string and `value` as an array in the payload.
13:50jgrahamA 1337 solution indeed
13:51AutomatedTesterLOL
13:51jimevansheh. not proud of it. but it works...
13:51jimevanslocal-end hacking ftw!
13:51AutomatedTesterawesome
13:52AutomatedTesterjimevans: could you raise a coordination bug for Selenium on this?
13:53AutomatedTesterala timeouts bug
13:53jimevanscan do. will take me a bit to get that on the books. i&#39;m slogging through my implementation errors right now.
13:53AutomatedTesterjimevans: no rush, I am running from meeting to meeting today
13:54AutomatedTesterif I get to it first I will let you know
13:54AutomatedTesterjgraham: that PR is merged
13:55jgrahamAutomatedTester: Thanks!
15:11jimevansanyone know if geckodriver understands get/setWindowRect yet?
15:11jimevansagain, lazy jim is lazy
15:14AutomatedTesterjimevans: not by the looks
15:15AutomatedTesterhttps://github.com/mozilla/webdriver-rust/blob/master/src/httpapi.rs
15:15jimevanswell, that&#39;s too bad.
15:16jgrahamI don&#39;t know if marionette supports it
15:16jimevansbecause there *is* no easy, hacky way around that
15:16jgrahamIf so it&#39;s trivial to add, otherwise it&#39;s not
15:16jgrahamato would know but I guess he&#39;s somewhere far away
15:17jimevansdammit, ato!
15:17jimevans:)
15:17AutomatedTesterjimevans: it&#39;s not in https://dxr.mozilla.org/mozilla-central/source/testing/marionette/driver.js
15:17AutomatedTesterso no, we havent started it
15:18AutomatedTesterjimevans: it should be trivial to add
15:18jimevansfair enough.
15:18AutomatedTesterjimevans: I will raise a bug
15:19jimevansi&#39;m going to update my local end to use it. window manipulating tests will fail against geckodriver.
15:20jimevanswhen geckodriver was the only remote-end spec implementation, it made sense to adhere to whatever it did. that statement is no longer true. :)
15:21jgrahamYeah, that&#39;s fine
15:22jgrahamFailing tests are more helpful than tests that hack around our bugs
20:13jimevansin geckodriver, if i execute script which raises an alert, does the executeScript call get blocked?
20:35AutomatedTesterone way to find out
20:43jimevanswhoa.
20:43jimevansnightly doesn&#39;t navigate to a url?!
20:47AutomatedTesterthat would surprise me
20:47jimevanssomething&#39;s really screwy here.
20:49jimevanswell, imma update my vm. then we&#39;ll see if a reboot fixed it.
20:50AutomatedTesterI will test in a sec
20:50* AutomatedTester is currently building firefox
20:51jimevansit&#39;s gotta be a local environment thing for me. i know this was working not long ago, and it was failing with the same behavior on release channel.
20:54AutomatedTesterjimevans: using Tip its working fine
20:54AutomatedTesternothing has changed recently in that area
20:55jimevanslike i said, i&#39;m quite sure it&#39;s something local.
20:55jimevansi&#39;ll take another look once this windows update is done.
20:59AutomatedTesterok
21:58jimevanswell, this is damned odd. i can&#39;t get anything to work with firefox and geckodriver.
21:59AutomatedTesterjimevans: I know that it works normally
22:00jimevansit&#39;s like the driver loses connection with the launched instance of the browser.
22:01jimevansoh. hang on.
22:03jimevansokay, i&#39;m an idiot.
22:03jimevansfigured it out.
22:03jimevansthat&#39;s what i get for not merging changes into my branch.
22:04jimevansi knew it&#39;d be something local on my end.
22:09jimevansand to answer the original question, no geckodriver doesn&#39;t block on the alert.
16 Mar 2017
No messages
   
Last message: 8 days and 4 hours ago