#selenium

depends if I wrote it

usually I CAN FORGIVE YSELF

whoops, caps

xD

don’t drink and chat folks

@simonstewart thanks for helping with merging that PR.

hazmeister-sl lol. Precious lessons :thumbsup:

NP. It pointed me to some of the grid code

And now my day has derailed a little

@simonstewart I sincerely from the bottom of my heart hope that I don't cause you to shed tears when looking at my PRs :)

Your PR is fine :)

I will go home now and work on the next PR. Already found my next GitHub issue to work on.

Phew! That's a relief @simonstewart

I'm normally pretty good at letting people know when I think we can improve something

My only hope is that I do that in a gentle way

[13selenium] 15shs96c pushed 1 new commit to 06master: 02https://git.io/fNaL9

13selenium/06master 149669e80 15Simon Stewart: Move utility method into utility class

Alright. When the java tests go as green as we expect them to, is there any reason not to ship 3.14?

go on!

AutomatedTester: ship your python diff and we'll move from there

Rubber duck time!

So, in my head the grid is composed of a number of moving pieces:

(Starting from the backend)

1/ end nodes (probably chromedriver, geckodriver, or similar)

2/ possibly a wrapper for that (eg. docker, saucelabs, browser stack)

(It's possible that Selenium As A Service should treated as "1"

)

3/ "Something" to route requests to the right destination

4/ a "distributor" which is aware of all nodes in the system, and is responsible for assigning sessions to nodes

So, "1" equates to a "session", "2" to a "host", "3" to the router, "4" is the distributor

We might want to put proxies in there to handle things like protocol conversion

Or we might want that to a responsibility of the session

In a k8s setup, you'd probably have a load balancer somewhere in the mix

(which is why I need to update the webdriver spec to make session cookies A Thing)

So you'd end up fronting all this with haproxy or something similar

And have leadership elections for the distributor

With a permanent datastore for the session routing

@tomer, yt?

I am

Whats up?

Two things: sizzle, and an IDE Alpha blog post

Alright

Seems like a sizzling topic

Ha! :)

So, would you like to nominate someone to write a blog post announcing the alphas of IDE?

I guess it'd need "where to get it", "how to use it", "what we want before we hit beta"

I can do that with @tourdedave

If there's nothing to do before "beta" (I think there is ;) ), we can go straight to announcing the beta

Lovely.

Ping me when you've something written and I'll make sure you have the keys to post to the blog

tourdedave may already have them....

Depends on what we call beta?

I should already

"beta" == "all the major features, including export, are in place, but we don't know how well they work"

"alpha" == "it may or may not work, and we know there are gaps"

gotya

"release" == "any bugs we document are features"

:)

how crucial is export?

It seems to be the thing that most people are wanting

That depends on the runner’s reception

If that could even give an answer to 50% on day 1

And we can make it more flexible as we go to cover almost all of the cases

Even if the runner was super, you'd still want to export to (say) C# or Java

We can reevaluate

Well, some might still want to

Indeed, but there are more pressing features

Because IDE is a step into the world of Selenium.

tomer-sl: that's why we announce an alpha

Right, but if we ultimately enable people to do everything in the IDE that you can do in code, then there's no real need to export

Uhhh...

I'm not sure I agree there.

i would disagree at 'enable people to do everything in the ide you can do in code'

thats just not going to be possible without allowing people to write code for the ide

One of the use cases for IDE is "I need to throw a quick example together". That doesn't imply people will actually be running the tests in IDE

I'm not saying "IDE should be rubbish", but I am saying that it's often an "on ramp" to the other bits of selenium

simonstewart: I get what you say, I’m not inclined to rewriting each feature 3-4 times atm

Hence "alpha"

Wanna talk about sizzle?

Sure.

So from a quick check I’m the culprit

I think Jason Huggins is, to be fair

It was disabled I reenabled it a while ago

So, stepping back

Sizzle was how RC used to do element finding if you asked for "css" locators

It was disabled at some point

So any test that uses the RC APIs and css actually depends on sizzle

They just failed then? until I reenabled it?

I need to go read the code for the webdriver-backed selenium

I want to flesh out the locators in the IDE

But, yeah, it should be there for backwards compatibility

The original IDE code was a hack on the original Selenium Core code

css and link are in my sights

Which is how we ended up with sizzle where we are now

tomer-sl: you should be able to use the atoms for locating elements

I do

I don’t need sizzle

Right

Since I incorporated that

But I need other things from there

So it's cool to just cut that entire chunk out for IDE TNG

Which other things?

getText and the likes

getText is also an atom

I’ll have to go over it

iirc I’m using over 17 utils functions all over the place

bot.dom.getVisibleText

RC's getText is hopelessly broken

Is it documented somewhere

bot.dom.getVisibleText?

I’ll have to check if I can port entirely away from it

And then I can just cut it from the build process

What are you using in the htmlutils file?

Easy to ask, it puts everything on window

I’ll have to grep

Thanks

one sec

const eval_css = window.global.eval_css; const getTimeoutTime = window.global.getTimeoutTime; const SeleniumError = window.global.SeleniumError; const normalizeSpaces = window.global.normalizeSpaces; const getText = window.global.getText; const createEventObject = window.global.createEventObject; const triggerKeyEvent = window.global.triggerKeyEvent; const fnBind = window.global.fnBind; const browserVersion = window.global.browse

window.global.getInputValue; const extractExceptionMessage = window.global.extractExceptionMessage; const PatternMatcher = window.global.PatternMatcher; const parse_kwargs = window.global.parse_kwargs; const getTagName = window.global.getTagName;

It's likely the atoms or modern JS already have replacements for almost everything

That’s a short list of things

OK. So "getText" we've covered already

bot.actions can do key events

const objectExtend = window.global.objectExtend; const SeleniumError = window.global.SeleniumError; const createEventObject = window.global.createEventObject; const absolutify = window.global.absolutify; const addLoadListener = window.global.addLoadListener; const highlight = window.global.highlight; const _parse_locator = window.global.parse_locator; const eval_xpath = window.global.eval_xpath; const eval_css = window.global.eva

getting a tag name is trivial

window.global.PatternMatcher; const getText = window.global.getText; const getTagName = window.global.getTagName; const elementFindFirstMatchingChild = window.global.elementFindFirstMatchingChild; const eval_locator = window.global.eval_locator; const is_ancestor = window.global.is_ancestor; const canonicalize = window.global.canonicalize; const parseUrl = window.global.parseUrl; const reassembleLocation = window.global.reassembl

window.global.serializeObject;

That’s everything

serialize sounds like JSON.stringify to me

IDE had some odd behaviour about looking through nested frames for elements which was always "fun"

And damnably confusing