Skip to content

Latest commit

 

History

History
186 lines (119 loc) · 6.41 KB

comparison.md

File metadata and controls

186 lines (119 loc) · 6.41 KB

Comparison

Far from being finished....

Intern web site contains some comparisons of the frameworks, however it is missing some important players and features. So I will add some here:

Subjective Rankings:

3-good
2-ok
1-poor
0-none
feature selenium-webdriver kommando intern
Standalone Server 1 3 1
Documentation - Local run 3 3 1
REPL 0 2 ?

Standalone Server

How much manual work is needed to run tests against browsers that require selenium-server-standalone-*.jar. Download, configure and instantiate support in each 'framework'.

selenium-webdriver - 1 -

  • (cons) 'To run the tests, you will need to download a copy of the ChromeDriver and make sure it can be found on your PATH.'

  • (cons) They don't event updated docs (nor npm nor wiki) after implementation of direct Firefox support.

  • However after some downloading of related and compatible version for each of your OSes, you can have it running in lt; 30 mins for all browsers you need.

    TODO: download sample code sample

intern

  • why using pretty old 'selenium-version': '2.41.0' in the example config ?
  • based on this blog post you must download all manually as well

kommand/vebdrvr

Actually 'installing' Selenium, Chromedriver, IEDriver and PhantomJS is not node by kommando, but by vebrdvr module.

Grunt and other tasks

to automate download of Standalone Server and Drivers

TODO:

Documentation - quick start testing with local browsers

Here I ry to evaluate how much time and browsing it took to start first functional test in local browser. Intern is in disadvantage here, their docs cover bit more so orientation is more difficult.

selenium - 3 -

running sample from npm readme with native FF support they have now, took under 10 seconds, if you have FF installed.

kommando - 3 -

Running REPL and starting browser (FF, chrome, Opera) right after install requires no reading of any docs. This is the main purpose of the kommando module (even if implemented using vebdrvr).

$ ./node_modules/.bin/kommando --runner repl --browser chrome
Starting Selenium server ...
Selenium server started at: http://192.168.x.x:49408/wd/hub
Run tests using "browserName=chrome"
"kommando" / "kommando.browser" provide access to the running browser instance
kommando>

intern

After 30 minutes of quick searching, I have found this topic: Running-your-own-WebDriver-server. After another 30 minutes I have still no functional testing sample running. Another 30mins, cloned tutorial, downloaded full tutorial cannot make tests/functional to run, no errors, just not run.

Gave up, asking for help

Update: 2015/02/10 - reading the answer (Thank) and googling I have found this http://www.sitepen.com/blog/2014/04/23/dojo-faq-how-can-i-run-dojo-tests-locally-with-intern/ article. After not believing this is really what intern offers, I have 'managed' to start server MANULLLY with actually typing the word 'java' (GRRRRRR, just compare wit the bare selenium-webdriver, which 'they' do not consider to be a test framework , and after previous positive KARMA, seems unbelievable achievement..

After "pruning my environment" (as vaguely suggested in article), to my guessed minimum:

{ browserName: 'firefox' }

or

{ browserName: 'chrome' }

and having test-code:

define([
	'intern!object',
], function (registerSuite) {
     registerSuite({
        name: 'minimal demo',
        'of getting google': function () {
            return this.remote.get('http://www.google.com');
        }
    });
});

and running:

(GRRRRRR) java -jar $SELENIUM_SERVER_JAR -port 4444

and

node_modules/.bin/intern-runner config=tests/intern

I have finally seen 'flash' of google.com webpage , after series of blinking URLs.... Success withing 48 hours in not bad. Or ? Going to sleep (today).

Coverage

Mocha

asking for help

Finally I have decided to ignore mochas docs and tj's repo and tried very simple one minute setup and example:

npm install --save-dev jscoverage

mocha -r jscoverage -R spec tests/mocha/test.js

Coverage result
  ✓ tests/mocha/test.js	line[100%]  branch[100%]
  ⁍ lib/module.js	line[ 66%]  branch[100%]

  1 passing (5ms)

After trying tests on my other Win (MINGW) machine, neeeded to quickly patch jscoverage.

And also received link from boneskull about blanket, so will try it.

Blanket

Testing blanket on windows (MINGW) ends with no conevarge stats at all, quick look at the node+mocha detection reveals why:

if (args[0] === 'node' &&
    args[1].indexOf(join('node_modules','mocha','bin')) > -1 &&
    blanketRequired){

    //using mocha cli
    module.exports = blanketNode(null,true);

So probably I'm going again for Fork-reconfigure-package.json-Fix-Pull-Request-Hope cycle ? Same as with jscoverage ?

Not today ;-(