#selenium

is there a way to use xpath in a more flexible way? for example, selenium ide generated the following code: "driver.findElement(By.xpath("//input[@...;, but i have multiple dropdowns that could dynamically contain the same 'Filtering' valued input element on a page reload. i'd like to do something more like this: "//*[@id='filter-site']...//input[@...; where '...' could be anything but is still contrained to

the 'filter-site' specific dropdown. is something like this possible?

you can put wilds between slashes anywhere in the xpath. so like "//*[@id='filter-site']/*/input[@value='Filtering']"

kyoob, oh wow, i was thinking that wasn't possible :).

better yet, use css: "#filter-site input[value='Filtering']"

^^ +1

jleyba: +1

i actually just used xpath for the first time in many many moons today

i have a known string and wanted to make sure it's in a <span>

kyoob: xpath has its uses. those uses are far fewer than most people expect.

totally

thanks. will try the css version

how outdated is this?: https://www.simple-talk.com/dotnet/.net-framewo...

"#filter-site input[value='Filtering']" doesn't seem to work in selenium ide

are you sure it's looking for css and not still xpath?

kyoob_, looks like jleyba's example uses both

"#filter-site input[value='Filtering']" is a css selector

no xpath in it

oh. shouldn't it still work in selenium ide? i don't see a way to specify whether it's xpath

css="..."

ahhh

cool. thanks

it's been a while but isn't the argument formatted like .. css="#filter-site input[value='Filtering']"

just another reason i abhor selenium ide.

now now

i'll eventually get away from it once i get these selectors down

selenium ide taught me perl

in a way

the export libraries are clutch fora a novice

simonstewart leaved the chat because you talk about selenium ide

ha

man, record and playback tools get a bad rap

kyoob: they're useful in some cases, but, like xpath, that utility is far more limited than most people really want to believe.

how should i structure Page Objects if i have a lot of javascript strangeness going on? for example i have five dropdowns i need to open/search/select/deselect items from but nothing changes between them other than that css id. would hate to write a custom method for each action with each dropdown.

For some reason dropdowns don't play nice with firefoxdriver after a parallax

eindoofus: private helper methods that take the element in as an argument?

the whole page shifts when attempting to select the dropdown

jimevans this is true. i think if someone could build a really super-capable selenium ide type thing, a lot of automation people would be out on the street.

kyoob: lots of people smarter than me have been trying to create something "really super-capable" for at least the last two decades. i've yet to see it.

dangit

kyoob_: you didn't miss much. just me saying "i've yet to see something 'really super-capable' in that space"

jimevans well, who would want to work on that project?

jimevans, so i would pass the element itself? i was thinking of passing just the css id as a sting and then having a method that concatenates it before grabbing the element within the helper method

but i've seen enough snake oil salesmen try to say they had it that i'm immediately skeptical

string*

No More RC!

For Science!

( ¯\_(ツ)_/¯ and On my way! are the other keyboard shortcuts i have too)

eindoofus: either way. the point is that you don't write the "custom method for each action with each dropdown". you write the custom method once, and execute the same operations a dropdown

of course, you haven't said whether you have page objects or page wrappers, but that's probably immaterial to this discussion.

jimevans, thanks. makes sense. what have you seen people do in terms of the method argument? are they passing elements or strings around. i don't want to do anything that is too hackish and deviates from good practice

jimevans, just Page Objects at this point

i don't think it matters which way you go as long as you're consistent throughout your project

eindoofus: so you expose no individual screen elements in your object classes? in other words, on a theoretical login page, you do not have "setUserName", "setPassword", and "clickLogin" methods?

jimevans, i'm confused. i do expose those as public methods. i basically implemented the example at the bottom: https://code.google.com/p/selenium/wiki/PageObj...

that's a bad example imo.

but whatever.

it's not that important.

uh oh. to what degree is it bad? that's the way i've been kind of going about it

been*

eindoofus: you're doing it the way most people do.

like i said, don't worry about it. it's not important.

please let me know if you ever write that blog. lol :)

i'm feeling like being a pedant about page objects vs. page wrappers.

a page wrapper basically includes methods to manipulate each element on the page, and little else.

as long as you get logged in somehow, right?

page *objects* on the other hand, model the services provided by the page. on a login page, the service would be logging into the application. so it would have a "loginAs" method. or a "submitIncorrectCredentials" method.

*NOT* methods to manipulate each individual element.

most people implement page wrappers, and (mistakenly, imo) call them "page objects"

kyoob: logging in through the UI is usually an antipattern for everything except explicitly testing the flow of the login pages themselves.

i was just going along with the example, is all

i know, but it needed to be said. :)(

For some reason, my chrome is crashing after 50 or so tests

can't seem to find the reason why

on windows?

yes

Windows 8.1

check your tasks, you might have a lot of hanging little chromedrivers

jimevans: I'm not sure I'm following your pattern. A Page Wrapper has methods to interact with username and password text fields. Are you saying a Page Object is a separate abstracted object to login using the elements defined in the Page Wrapper?

sometimes i close out a lot of chromedriver tasks and my tests run like a dream. they seem to open when i'm debugging in eclipse fwiw

titusfortner: i'm saying that you use one or the other. internally, a page object might resemble a page wrapper, it just doesn't expose the methods to interact with username and password fields outside itself.

but no, it's not a separate object.

or needn't be

i thought that was the problem too, but I only have one instance of chrome river going right now

i suppose you *could* do it that way. i wouldn't.

cweng: then i'm stumped!

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

13selenium/06master 14763bbcb 15rbri: update to htmlunit to 2.16...

[13selenium] 15lukeis closed pull request #496: Update to Htmluni 2.16 (06master...06htmluni_-2_16) 02http://git.io/vfQaZ

03rbri closed Update to Htmluni 2.16 - https://github.com/SeleniumHQ/selenium/pull/496 []

cweng: Are you quitting the chrome browser or just closing it?

also are you spinning up a separate browser for each test?

kyoob: hahaha, same thought. :)

I am suppressinging the GC until all tests are run

using GC.SuppressFinalize(this);

I'm quitting, closing, and disposing the driver each test as well

cweng: I'm admittedly pretty new to this, but chrome typically runs like a memory hog. I'm inclined to wonder if chrome is just crashing because it's too bloated.

ah, never mind then.

how does it crash? 'chrome not reachable' error?

Operation has timed out

HTTP request to the remote wevdriver server timed out after 60 seconds

And then another browser would't be opened

I'm getting an error on my selenium server when trying to run a chrome test saying the path to driver executable must be set. Does this get set on the server or client?

testerbit: set by you

automatedtester: set by me on the client-side or server side?

testerbit: whereever your chrome is

the error I'm reading off the server

chrome is on the server, where would I define on the server?

testerbit: you need to add it to the PATH variable

so |export PATH=$PATH:/path/to/chromedriver| for *nix machines

i added the directory to the path, this is a windows machine

that should work

testerbit: we normally just try run chromedriver

as long as the path isn't under \windows\system

does not work

Firefox and IE drivers still work, only chrome crashes after 40 or so tests

"The path to driver executable must be set by webdriver.chrome.driver"

testerbit: set the property then

Newbie question. Using C#/NUnit/Selenium. When you start a test in NUnit, where does processing begin exactly?

I mean where in your code.