#selenium

`WebDriverException: Message: Parameter 'duration' was not a positive integer`

ha

that's strange since it's supposed to be multiplied...

simonstewart: barancev: The custom RemoteWebElement class "wraps" calls to findElement to find an element that could exist in a nested frame(s)

kpschmid_: you walk deep through all nested frames and their subframes and loko for elements with the given locator?

*look

ah that's a bug...

barancev: correct

kpschmid_: well, give me some time, on Monday I hope to create a wrapper around public API that implements this feature

barancev: Did you look at my gist?

kpschmid_: yes, I just wanted to clarify the intention

barancev: This approach of overriding JsonToWebElementConverter.newRemoteWebElement() has worked very well for us for several years

because you also told about "gui maps", it's a different feature

barancev: Ignore the guimap stuff, not looking for a "new" feature, just have the override working again

kpschmid_: simonstewart is working on the fix :)

kpschmid_: but I'm going to suggest you an alternatve solution based on public API

barancev: I'm a little confused as to how overriding JsonToWebElementConverter.newRemoteWebElement() is not "public"?

it's protected, at least :)))

It *was* changed to private in 3.4.0 and then changed back to protected in 3.5.x

hope simonstewart fill fix it

Working on it

but I'd suggest to deprecate it and remove soon :)

simonstewart: barancev: understand

it was a *bad* idea to implement this injection hook

barancev: Unfortunately, the person that implemented our framework leveraged it dating back to 2.x and looks like some extensive refactoring to "un-do" this ditty

[13selenium] 15natelowry opened pull request #4624: fix typo in docs for driverExecutableFileName (06master...06patch-1) 02https://git.io/v5RU0

kpschmid_: I mean us, selenium devs, bad idea to provide this hook, to make it available to end users

apailthorp-sl: it appears there's lots of multiplying going on in the python bindings... which is why you're probably seeing it sit there for a long time

[13selenium] 15shs96c pushed 2 new commits to 06master: 02https://git.io/v5RTv

13selenium/06master 14d80f86d 15Simon Stewart: Revivify the ProtocolConverterTest

13selenium/06master 14c8f6bf9 15Simon Stewart: Move element converter so it's only used in RemoteWebDriver...

@lucast strangeness abounds, getting hangs in places I don't think much is going on in. Not feeling warm and fuzzy about geckodriver.

barancev: I bet it made sense at the time

sure, all decisions have resons

the pause issue would be python's fault (at least until i diagnose it further)

going to let this thing sit while I go get a burrito. Maybe it will throw a useful error if I wait long enough and spin in my chair three times.

Pushed the fix

Hi, does anyone know if it's possible to reconnect to a previously existing remote webdriver session?

I've been looking in the group and a lot of googling but I can't get a straight answer

simonstewart: Thank you

adamp, I don't remember if it was you or someone else, but yes it's technically possible (I don't remember how) but it's not advisable to do AFAIK.

ok, I've been able to get it working with the python bindings by setting driver.session_id however in the java world the set session functions are protected - so yeah it doesn't look like they're meant to be exposed

adamp: if you know the session id it's possible

yeah I have the session id, that's easily available. I'm just not able to use it to force a remote webdriver to latch onto the existing session. Are there any java api examples out there does anyone know?

I can see there's a setSessionId function: setSessionId(String) - Method in class org.openqa.selenium.remote.RemoteWebDriver

but it's protected not public

apailthorp-sl: i'm not getting any hanging issues that you were

what version of geckodriver do you have?

@lucast 0.18

64 bit on Ubuntu

same here

not consistent, hung near the end of test suite before lunch, now hung at start. If I break and restart betting I will get another example, but different.

I wonder if it's safe to turn the AlertsTest back on

kereliuk: https://bugs.chromium.org/p/chromedriver/issues... is blocking that

apailthorp-sl: can you try this?

[13selenium] 15shs96c pushed 1 new commit to 06master: 02https://git.io/v5Rtr

13selenium/06master 144cacc9d 15Simon Stewart: Add back test ensuring preservation of pointer location...

apailthorp-sl and lucast: you'll need to add the pause to the keyboard events too. Apparently there are two lists to track

@lucast sure right after this status meeting...

simonstewart: I'll check up on the status of that issue

Thank you

Keyboard events? Not following.

simonstewart: not if there are no keyboard events :)

I beliebve it'll send both sets of events

it'll still send some pauses for keyboard, but since im not using it they dont matter

now if i tried to send some keys afterward.. that'd be an issue

[13selenium] 15juangj pushed 1 new commit to 06master: 02https://git.io/v5RmT

13selenium/06master 14b5999f6 15Jakub Vrána: Always use JSON.parse for bot.json.parse (#4603)...

@lucast that all seemed to go well, output is https://gist.github.com/980b9fb0d6aeae121370ace...

Not sure how many folks would be keen on learning more stuff, but this seems like an awesome deal! https://deals.gdgt.com/sales/pwyw-learn-to-code...

Not sure why yours isn't working then

Thanks in any case, appreciate the consideration

[13selenium] 15jimevans pushed 4 new commits to 06master: 02https://git.io/v5RsB

13selenium/06master 14f3208bf 15Jim Evans: Changing .NET ChromeOptions to use goog:chromeOptions as property name...

13selenium/06master 145638427 15Jim Evans: Simplifying Proxy properties marked as obsolete in .NET

13selenium/06master 1474fb803 15Jim Evans: Moving WebDriverWait into main WebDriver.dll assembly...

https://github.com/SeleniumHQ/selenium/wiki/Chr... | http://chromedriver.storage.googleapis.com/inde...

[13selenium] 15shs96c pushed 1 new commit to 06master: 02https://git.io/v5RlJ

13selenium/06master 1407b7b54 15Simon Stewart: Replace cglib with bytebuddy...

webpage says that an element is at (13,683) but when I try clicking it, the page outputs that the click was at (37,695)

Anyone have tips on why or how to resolve this? (aside from manually adjusting the numbers?)

In my last test I clicked two different times and it seems like the y access was the same amount incorrect. 683 vs 695

Though, I suppose that may not matter as on click events for the buttons are outputting text. :grumbles off to work on it some more:

js-sl any animation involved?

Absolutely. :P

Now that I'm not chasing click location issues, it's likely apple's screwery around video playback.

I just solved this on iOS 10 but now trying to get it solved again on 7-9

geeze iOS7?

what are your usage numbers back there?

I don't have specific numbers. lol. I probably shouldn't even both with it tbh. Apple says that <9.0 is 3% share

we mainly stick with current and previous iOS versions

Right now I can't even get this resolved on 9 though :P

so we support 9 & 10 now

you using Appium?

II work for an online video saas platform, so our customers could have stupid requirements.

Yeah, through Test Object.

I there was a bug with locations and viewports I raised a while back

I'm testing mobile web in Safari

you might need to use javascript to find the position

Maybe, but I've got this outputting text appropriately:

So, that's got me thinking it's more than a misalignment there