What does your test look like? As in, what does your assertion look like?

We don’t test at that level. Just “user visits this page, does this and that, and then this should be on the screen”.

But you could, using JS execution.

@raymcdermott

I can’t add code snippets here

I’ll put in the main page

ah ok

or msg me privately, also fine

I think the discussion is OK in the forum - no?

I think so.

what code?

the js code with scrolltoview

maybe it was just a sample of what was possible?

ok, it’s just showing how you can execute JS. In your case it would be (eta/js-execute “(whatever-test)“) and then assert from the JVM that everything is fine

FWIW, I’m using this Docker to spin up a headless chrome: https://hub.docker.com/r/robcherry/docker-chromedriver/

thanks

but during development I’m not using a headless one btw, I want to look at what it’s doing

how does the JVM report on the tests?

re non-headless … yes, sure

well, you can use regular clojure.test to drive etaoin. e.g. https://github.com/igrishaev/etaoin#writing-integration-tests-for-your-application but I’ve written my own glue between clojure.test and etaoin

you can do e.g. (is (= 3 (eta/js-execute "1+2"))), but I’m not sure how far you would get with that

I’m getting confused about why I would use js-execute

the other stuff all seems nice / legit

to execute your re-frame test inside the browser

so I must be missing something obvious

I am assuming this code runs inside a browser where your app is loaded, no? https://clojurians.slack.com/files/U04V5V0V4/F8RCVN0JW/re-frame-tests.clj

oh, I see - use that instead of Karma

sorry, I was a bit slow there

right. but again, I’m not sure how nice that would be to work with. np

ok, thanks though … I now get that it’s an option

though does look a bit yucky tbh …. but then everything has some yuck on it 😉

it’s the law of conservation of yuck

haha

btw are you going to clojured?

I’m hesitant, because it’s only one day… the travel would be longer than the fun

I know what you mean …. but we’re having a defn road trip … both Vijay and myself will be talking

I’m looking forward to the report in one of the next defn episodes!

haha yes we will be sure to do that

btw, the etaoin integration test is the first moment that I have used a dynamic var!

yeah, I’m also doing that

scary 😱

actually in tests it’s the only place where I use it

yes, should be safe

test code is more yuck-accepting

yuck-forgiving

yuck-creating

we need a yuck management system

oh yeah - clojure 🙂

Clojure is the lotus built upon the mud

well said

definitely coming to Dutch Clojure Days though

likewise … so see you there

should be a vegetable weekend to remember

betacarotene overdose

friendly reminder - this is a not a porn server

and the slack emoji set does not include a carrot!

who’s thinking of porn here then?

😇

betacarotene is my safe word

Btw, I hope you read the CoC (code of conduKt) for ClojureD?

lol I’ll try to keep it clean

but I see what you did there

Reinheitsgebot is a thing in Germany

these are the marks of a truly civilised society

no chlorine in the beer / chickens

quick question ….

assuming you have a login button

how do you test for it’s visibility?

(I can’t see a way to provide an ID in reframe)

I have tried {:class :rc-button.btn.btn-default} without luck

<button class="rc-button btn btn-default" style="flex: 0 0 auto;">Login</button>

the class has spaces so I don’t know how to transpose that

do you mean from etaoin?

yes

I do it like this:

(defn contains-text? [text]
  (format “//*[contains(text(),‘%s’)]” text))

(defn contains-class? [class]
  (format “//*[contains(@class, ‘%s’)]” class))

and then e.g.:

(eta/wait-exists *chrome* [{:tag :div :class “search-summary-text”}
                                 (contains-text? “results”)])

ok, nice …. thanks, I’ll give it a shot

note that when you use :class it only matches on the full class, so this is a workaround for it

There’s also https://github.com/igrishaev/etaoin/blob/master/src/etaoin/api.clj#L1662 but not sure where to use that

hmmm doesn’t that effectively replace your contains-text function?

Not sure about that

chasing the code, it runs query which seems to include matches like you have in your fn

A query might be: - a string, so the current browser’s locator will be used. Examples: //div[@id=‘content’] for XPath, div.article for CSS selector

no?

anyway … I’ll give that a shot first and see how it goes 🙂

I gave you an example with a query.

[{:tag :div :class "search-summary-text"}
                                 (contains-text? "results")]

has-class seems to do the same thing yeah, has-text is also there. It didn’t exist when I wrote this. I’m an early adopter 🙂

haha well at least it means I’m not cracking up

cracking up?

going crazy

aah

sorry I’m an idiom idiot

well, I learned English mainly by watching TV, but it only goes so far

I tried to learn Flemish by watching TV but I can only endure one form of punishment at a time

haha

it’s actually even simpler …

this works (wait-visible {:class "rc-button btn btn-default"})

yes, that also works, but you’d have to specify the full class. also visible can cause unexpected results, because in some cases the element never becomes visible. That’s why I use wait-exists instead

visible depends on the viewport your using

sure … but in this case it must become visible (login) or we are screwed 😉

but I take your general point

yeah, sometimes the user has to scroll a bit

working with the full class works, until you add a class… e.g. “active”, then it quickly breaks

yeah, this is baby steps for me

and babies fall over all the time

but it’s exciting 🙂

yeah, it takes a few babies to get there, but it’s worth it

haha

thanks for picking me up off the floor and getting me walking again

lol

The nice thing about etaoin is that it’s just a simple wrapper around an API, so if something is missing, just add a clj-http call somewhere

much better than Selenium Java crap if you ask me

Thanks for the heads up via Patreon 😛

?

Oh, I thought you sent it

but then it must be Vijay

sent what?

https://www.dropbox.com/s/ubzkz85943zgsjk/Screenshot%202018-01-12%2012.07.50.png?dl=0

LOL yes he just informed me in another window

if you have a question about the impact of post-Hegelian idealism on Clojure, now is your chance

am trying to do an oauth2 flow using auth0 and it all works fine with chrome but fails when running headless …. have you seen anything like that?

[ I can open an issue of course but just wondering ]

maybe it’s the visible/exists issue?

I played with that in one or two steps without success

but let me check

I’d also make a screenshot for debugging

+ you can execute some arbitrary js to test for elements etc. for debugging from the REPL

seems like the auth0 overlay has to be visible

REPL your way through it

I have used the REPL to build the script visually … I live the dream

How do Haskell people even get by. The type system doesn’t help you in any way with issues like this.

I mean any language without a real REPL

Don’t have anything to ask Zach, except to thank him for Aleph, the fundamental thing in Yada that we are gratefully using

agreed on REPL

you have time to think on Zach but yes, we are all grateful for aleph

seems like the google oauth2 form changes whether visible or not 😞

but yeah, amazing tools we have

didn’t understand

when running chrome normally I get a different login form (different field names) from the one I get when running headless

ah tricky. whyyyyy.

fuck knows

but nice tools so that I could discover it 🙂

being able to grab the html is killer

yeah