#selenium

morning guys

i would like to ask you about how do you show in console or log which step of a testcase is the test on

simonstewart: yt?

barancev|3: kind of

End of the day here

G'day jimevans!

re: issue #5064

03AndyUM357 open WebDriver v3.7.0 C# Firefox "missing handle parameter" for popup windows - https://github.com/SeleniumHQ/selenium/issues/5064 [C-grid D-firefox]

and it's :friday

barancev|3: what about the issue is troublesome?

sounds like cloud vendors expect bindings send W3C payload despite the fact they act like OSS implementations

Then that's a bug for the cloud vendor to fix

can we send mixed payload (both name and handle) at the same time?

We could, but it's really just a problem for the cloud vendor to fix

yes, I see, just to work around this issue

If we put in the workaround, the actual fix will never be implemented

true :)

And there are SauceLabs folk here

(Is it BrowserStack too?)

I think there are BrowserStack folks here too, but I could be wrong

according to the comments, yes

ok, got it, we won't fix it :)

s/fix/work around/

Though we might want to check that the protocol conversion does the Right Thing

what protocol conversion?

we use pass -hrough now

pass-thought

The Selenium Server does protocol conversion between w3c and jwp dialects

does it?

Passthrough only matters when the downstream and upstream ends speak the same dialect

Yes

:)_

there is no more option to disable passthrough mode

See above

and I thought it's the binding that should adapt to the protocol version

Passthrough only matters when the upstream and downstream nodes speak the same protocol dialect

The server handles it too

if it's our server

we don't know what saucelabs implementation does

I can't reproduce the issue with vanilla selenium server

We don't know

Ah! https://github.com/SeleniumHQ/selenium/blob/855...

03barancev 2 months ago - Deleting unused private field | https://github.com/SeleniumHQ/selenium/commit/8...

This is where the magic happens

barancev|3: yt?

here

consider the following element: <a>&nbsp;link text&nbsp;</a>

find by link text "link text"

atom won't find it

(but will find same item wrapped by normal breaking spaces)

question is, whether the atom should replace non-breaking spaces with normal breaking spaces (at least for finding purposes)

afaik it calls an equivaent of getVisibleText, that converts &nbsp; to regular spaces

apparently not.

let me look at the code...

it does, in fact, call getVisibleText, but the finder doesn't collapse the leading or trailing space for finding on full link text if they're &nbsp;

though it does collapse those for normal breaking spaces.

yes, it does not collapse them

it collapses regular spaces only

too sophisticated for friday evening :)

jimevans: I think the magic is implemented in bot.dom.concatenateCleanedLines_

first of all getVisibleText changes non-breaking spaces to \xa0, then collapses all breaking spaces, and then replaces \xa0 with regular spaces

which leaves regular spaces at the beginning and the end if they were non-breaking spaces in the original string.

yes

and what the spec says?

kinda seems like we shouldn't do that if getText() is supposed to trim spaces from beginning and end.

:spec

https://w3c.github.io/webdriver/webdriver-spec.... | https://github.com/w3c/webdriver | bugs: http://goo.gl/LxCtcV | tests: https://goo.gl/c0jYUV

the spec says that finding by (full) element text should be done be "removing all whitespace from the start and the end of the string", and "whitespace" is further defined as "characters from the Unicode Character Database with the Unicode character property 'WSpace=Y'".

1. Let rendered text be the value that would be returned via a call to Get Element Text for element.

2. Let trimmed text be the result of removing all whitespace from the start and end of the string rendered text.

3. If trimmed text equals selector, append element to result.

so, yes, we should change the atom

ah. but the question is, does the unicode spec consider 0xA0 to be whitespace?

in other words, does that code point have the "WSpace=Y" property set?

(java's Character class does not consider that character whitespace)

I don't know

i'm trying to figure out whether the test may be incorrect

(it's a new web-platform-tests test)

barancev: okay, the find element atom is what needs to change to be spec compliant.

let me look into that.

Hey guys, trying to do a Multiple File Upload using SendKeys with ChromeDriver in C#...

Is anyone familiar with a way to format the string to send to SendKeys?

ninsei: the .NET bindings really aren't geared toward uploading multiple files at one go.

with SendKeys