#selenium

Please read http://sscce.org/

I fear that at some point, I may have to contribute code to the php bindings

What is the tentative plan/timetable for deprecating support for legacy FirefoxDriver?

kpschmidt: I believe it will be after Firefox ESR-1 is no longer supported by it

yes, it's tied to ESR lifecycle

2018-03-05 is when ESR becomes 59.0 and ESR-1 will be 52.x

banrancev|2: https://gist.github.com/paulbriton/712bc325adc2...

davehunt-sl: thanks

Hi, I'm having an issue with selenium / webdriver / chrome on heroku, it looks like someone potentially solved the issue here: https://github.com/heroku/heroku-buildpack-goog... but I'm having trouble translating that to the appropriate ruby / selenium settings, could anyone help with that?

kpschmidt: When Firefox ESR-1 no longer supports XPI

simonstewart: I'm migrating from old phpunit bindings to facebook ones. I've asked the main dev of the project if he is interested to get closer to selenium team for better support

I may poll users to find out how wide-spread the old Firefox usage is at that point

pauluchon: It'd be nice to see them here

simonstewart: I'm still waiting an answer but he's quite active on github, I've proposed my help in order to implement the new spec in those bindings.

I'll ping FB too

Thank's

simonstewart: It is currently our fallback/stable grid solution. We have a 2nd grid in AWS running java bindings v3.4.0 but use proxies extensively.

simonstewart: also some of my test teams are reporting tests get queued in grid but never get executed but reported as fail.

kpschmidt: does that happen with 3.5.3?

this is for the AWS 3.4.0 grid... no problem with legacy grid running v2.53

I have not deployed 3.5.3 yet as I currently have an issue with proxies 'InvalidArgumentException: direct was not a valid proxyType value'

Interesting

Which driver reports that?

geckodriver

simonstewart: geckodriver for InvalidArgumentException that is

Interesting. geckodriver 0.18?

correct

AutomatedTester: any idea ^^

TL;DR: proxyType "direct" not being accepted by geckodriver

whimboo might have an idea too

I think moz#1387092 is it

03Henrik Skupin (:whimboo) RESOLVED/FIXED Add support for proxyType "direct" (noproxy) - https://bugzilla.mozilla.org/show_bug.cgi?id=13...

barancev|2: https://gist.github.com/paulbriton/712bc325adc2... Does it help ?

There you go, kpschmidt. Fixed in the next release :)

simonstewart: awesome! so it is a marionette fix?

Yeah

simonstewart: Can the Proxy object now be used with 3.5.3? Or is there work still being done in that area?

jasonmintz-sl: :api

.NET: http://goo.gl/uutZjZ | Java: http://goo.gl/Grc6tm | Ruby: http://goo.gl/jzh4RU | Python: http://goo.gl/sG1GfQ | Javascript: http://goo.gl/hohAut

ruby api is there ^ constructors for the driver should be in there

you’ll probably need to pass in the chromedriver path if thats the same issue youre having

Does anyone know if you can run Selenium nodes on Windows Server Core? (Trying to come up with a solution for testing IE/Edge given the infra we have)

tombee-sl: saucelabs use windows server 2012 to run nodes on windows with

not sure what Core is

I fear the “There is no Windows shell and very limited GUI functionality.” might be a problem

tombee-sl: I'm using a MS provided Win10/Edge VM as an AMI in AWS, but it is only good for 90 days :-(

pauluchon: rewrote to Java binding, can't reproduce

first click deselects an option

tombee-sl: it doesnt include IE/edge

so you might have issues there

:D

kpschmidt: the proxy object has always been usable with 3.x. There's been some churn in the spec that geckodriver has had to chew through, though.

tombee-sl: Edge is only available on Win10

Turns out, the end point implementation is harder than serializing a data structure to JSON

barancev|2: Thank's for your help. I've unselect with JS as a workaround. I'll fill an issue later on my bindings.

simonstewart: Yes, but it yields errors when I use it: 'InvalidArgumentError: Expected [object Undefined] undefined to be an integer'

In all drivers, or just geckodriver?

@kpschmidt: Ah, interesting.. alright.

simonstewart: geckodriver

(Imagine I'm editing the spec at the same time as listening on this channel)

OK, so chromedriver works?

simonstewart: correct

Or, better, IEDriverServer?

OK

it's a known issue in geckodriver

whimboo or AutomatedTester: when's the next geckodriver release?

there is a bug report

simonstewart: haven't checked IEDriverServer yet

geckodriver and IEDriverServer are the two current implementations of the w3c protocol

simonstewart: understand

kpschmidt: https://github.com/mozilla/geckodriver/issues/764

barancev|2: Yes, I can work-around it by creating jsonObject and splitting out host/port, but I cannot use Proxy object

@kpschmidt your modern,ie VM is on all the time?

or do you boot it when you need it?

I’m asking because I wondering if the licensing in Win10 flips out when the underlying CPU changes

barancev|2: correction, I'm 'network.proxy.*' in FF profile as current work-around

barancev|2: Do you have tested it with chromedriver ? (multiple select)

pauluchon: yes, chromeDriver 2.31

Ok thank's

hazmeister-sl: we have a few GA nodes any test team can use that have an associated autoscaling group, but all teams can run a CloudFormation template to create nodes for their specific team

hazmeister-sl: all out teams 'tag' AWS resouces, based on team, for billing purposes. ASG's allows use to dial up/down nodes as needed

OK, So I've been thinking about the about:blank technique for installing window.onload functions, but I don't know if I understood correctly. What would be the order of events? I'm trying .get("about:blank") and then .execute_script("window.location = ...; window.onload = function() { ... }")

I don't think my onload function is being executed

I’ve just discovered what ⌘ + K does on Slack

my life is so much better now

Jump to?

(On wind)

zavelevsky-sl: :O

toddpratt: im not sure the about:blank page is a real webpage

I asked myself why it’s so irritating to work when signed in to multiple domains

i remember something special being about it

luke-sl yep

[13selenium] 15vrana opened pull request #4603: Always use JSON.parse for bot.json.parse (06master...06patch-1) 02https://git.io/v5CQW

hazmeister-sl: The MS provided VM needs to be converted to an AMI so there is no underlying CPU change

do you terminate them when you are finished?

I’ve heard if you boot up a regular Windows VM on AWS, it might switch hosts and force you to reactivate

I'm using chromedriver and trying to install a window.onload function. I need to register backbone events

toddpratt: why are you trying to add onload functions to existing webpages?

toddpratt: what are you doing? Writing tests? scraping?

hazmeister-sl: no, we have teams testing at all times across mutiple time zones

scraping

toddpratt: ok, and why are you trying to add these onloads?

@kpschmidt thanks - I just started with switching to custom AMIs after using Server 2008

I need to catch events that are only triggered at page load

hazmeister-sl: for our non-Edge stuff (ie11, Chrome, Firefox) we still us community Windows Server 2016 AMI's and have no reactivate issues

I think the server version of IE causes some issues with text input

how are you finding ie11 on there?

I’m getting intermittant failure to send any text on server, and the odd missed character on Windows 7

hazmeister-sl: I'm not getting any issues reported from testing teams - we use the 32-bit IEDriverServer not the 64-bit

same here

thanks for sharing :)

hazmeister, you could implement through gems (or jars), some type of failsafe(rs)

so i knnow on our android VMs we have a keyboard class which basically uses `send_keys` and then Checks the keys sent.

hazmeister-sl: np

if they're wrong, it tries again. It's a bit clumsy, but it covers the 0.01%

toddpratt: even if you injected them on about:blank, they may not be there when you load a different webpage

:logs

hazmeister-sl: our testing it not bound to underlying OS, only browsers. We've been on Windows Server 2008 RC2 for 4-5yrs testing IE8-11