#selenium

the page I'm testing is a page that has parallax if that matters

g'day everyone.

howdy

hi

somebody needs to take the bacon and we need to push a release.

[13selenium] 15SevInf opened pull request #505: IEDriver's isDisaplyed should ignore opacitiy. (06master...06fix/ie-opacity) 02http://git.io/vJJwV

03SevInf open IEDriver's isDisaplyed should ignore opacitiy. - https://github.com/SeleniumHQ/selenium/pull/505 []

chromedriver is crashing after 50 tests using xunit, gonna try mstest

.25

i think maybe i'll use a power drill on my temple. it'd be just as painful as trying to make all users happy, but i might eventually come to enjoy that.

there's some contingent of users who expect elements with opacity=0 to be visible (isDisplayed == true), and some that expect those to be invisible (isDisplayed == false). it's simply impossible to make both sets of users happy.

jimevans: make it swap at 50%

comprimise!

[13selenium] 15jimevans closed pull request #505: IEDriver's isDisaplyed should ignore opacitiy. (06master...06fix/ie-opacity) 02http://git.io/vJJwV

03SevInf closed IEDriver's isDisaplyed should ignore opacitiy. - https://github.com/SeleniumHQ/selenium/pull/505 []

what happened to third_party/java/commons http?

nevermind.. it's apache-httpclient

damn its doing the same thing in mstest, so i think its a chromedriver issue

Unable to receive message from renderer

[13selenium] 15ddavison opened pull request #506: fixing intellij compilation issues brought on by 763bbcb4 (06master...06fix-intellij-libraries) 02http://git.io/vJJbC

03ddavison open fixing intellij compilation issues brought on by 763bbcb4 - https://github.com/SeleniumHQ/selenium/pull/506 []

03rbri 2 days ago - update to htmlunit to 2.16 | https://github.com/SeleniumHQ/selenium/commit/7... | http://dashboard.ci.seleniumhq.org/#/revision/7...

update chromedriver to 2.15 and it still fails on the 40th test

cweng: It may very be that chrome is executing too quickly. You may need to implement some waits.

consistently causes the chrome browser to freeze

I wonder whether it's worth ripping out apache-httpclient and replacing with okhttp

Probably worthless until htmlunit does the same

I'm sure this has been discussed / debated before… but when i was doing that pr, i was thinking… is there a specific reason we aren't using maven, ivy, or any other dependency management solution?

that pr = #506

03ddavison open fixing intellij compilation issues brought on by 763bbcb4 - https://github.com/SeleniumHQ/selenium/pull/506 []

[13selenium] 15jimevans pushed 1 new commit to 06master: 02http://git.io/vJJhx

13selenium/06master 14698dc42 15Jim Evans: Changing IE driver to allow interaction with elements where opacity = 0...

MonkeysPaw: Should I just add implicit wait to the driver then?

simonstewart: you wanna speak to sircapsalot's question? :)

Which one?

about dependency management solutions.

Because it's a fucking terrible idea

cweng: I've had chrome do the same damn thing to me. FF and IE will work fine -ish, and chrome will behave for several tests and then it will just crap its pants for no discernable reason.

So I'll tack in a wait.

Maven destroys your ability to actually manage your third party deps while pretending it's helping

I've been given the impression that explicit waits are preferable to implicit waits, however.

Downloading required deps from the web at build time destroys build times (natch) and increases build fragility

And means that someone who's frequently offline or has shitty connectivity can't do work

Until recently maven's public repos were only available over HTTP, leaving your builds open to MITM attacks

Maven dependencies are frequently poorly declared in poms

So, you know, just an already shitty "solution" to a problem solved by being an adult and managing deps yourself

Got to go and chat to people about status codes. Wish me luck.

good luck.

is there a way to wildcard version in cababilities? i tried '10.*', but that didn't work

simonstewart: maven was a huge improvement over life before maven, and assuming all your dependencies are already on your local repo you don't need to get any of them over the web

also, they can all be in a local repository on your local network

it is rather handy for managing multiple layers of internal builds as well.

the security issues are an interesting topic though.

the best thing about maven now perhaps though is that it is a standard.

it is rather self-evident though that i have different ideas about what is a best practice than the powers that be in this community, but what i found most frustrating was how difficult it was to load the java based grid code into an ide to debug it

when i asked others how they debugged it seemed to be by adding System.out.println() statements to the code and rebuilding it...

i am intrigued as to why people hate maven so much though :)

i dont

maven/gradle are great

im not sure what simon's issue is

I also <3 maven

not using maven doesnt exclude you from having to download new dependencies

indeed :)

and if you want to stick to a specific version you can easily specify it

maven gives u nice depencency graphs

dependency*

at least i'm not alone, thanks guys :D

I still don't see what the selling point of grid is though. Sounds to me like alot of pain for little benefit

yup

i didn't have so much pain with grid, and a lot of benefit

we had hundreds of vms running different oss, browsers, versions

if you don't mind giving me a few lines on it, learn me

mhall_work: grid allows you to have your actual testing run on a separate vm/pc than the CI vm/pc

and could run tons of tests in parallel via jenkins on all the browsers / vms / versions we wanted

mhall_work: it is especially useful if you have a huge CI system and need to run multiple tests at once

but even for a small CI system with one node it is useful

so I hadn't thought about the idea of running tests on something other than the CI slave in that way, that's interesting.

We use browserstack

in my experience, CI works best when all the tests can be run quickly, so that developers don't have to wait a long time after they commit to see if they broke anything

so thus, grid is moot for me

i want to check that out - that is essentiall grid, outsourced to the cloud

alko +1

we tried using saucelabs but it was so slow it was a joke

yeah saucelabs is saddening

we have a dashboard and up till now i've been throwing every method into the same DashboardPage page object, and every method happily returns the same DashboardPage object after I perform a method like searchDropdown(...). however, there is a part of the dashboard that is more complex and it would make sense to break it out if i can into another Page Object, however, i'm worried about loosing the flexibility i have now once i create a

DashboardSomeComponentPage object and enter that.

others are also using cloud providers like aws

eindoofus: have it inherit from the DashboardPage?

^^^

my 2 cents

when i get back home i am going to setup and run some benchmarks on cloud vm provider vs outsourced vs locally hosted solution performance

alko: Just open the selenium project in IntelliJ or Eclipse.

Everything works

conceptually, maven is great, but in practice it can lead to crappy situations

makes sense. but what do i keep in the DashboardPage object and what do i put in subclasses?

eindoofus: thats up to you

sometimes our project "breaks" because our dependencies are out of date/incorrect

if i am wrong then ok