Hi,

I experienced the same problem last day I was working, I remember that the problem was also in stable branches, where it should not be, as the hardcoded versions in composer.json prevents this kind of problems.

Requesting review, I will do the following after review:

1. Replace branches with moodle/composer.json tag change (for integration)
2. Pull changes on moodle-behat-extension.

For testing: please use attached composer.json.

Results for MDL-45698

• Remote repository: https://github.com/rajeshtaneja/moodle-behat-extension.git

• Remote branch dev-45701 to be integrated into upstream master
  • Executed job http://integration.moodle.org/job/Precheck%20remote%20branch/3585
  • Error: The dev-45701 branch at https://github.com/rajeshtaneja/moodle-behat-extension.git and master does not have any common ancestor.

Nothing that symfony/DomCrawler and other components are not mentioned here, as symfony's use same version for all components.
https://github.com/symfony/symfony/blob/master/composer.json#L26

Hi Raj,

About adding symfony, it makes sense and even more if all of it's components uses the same version, the browser one is probably the most used of them so if something have to fail will probably be there; IMO would be good to include in moodle-behat-extension/composer.json all the components that we are extensively using (like symfony or the web-driver one that I can think of now)

About hardcoding versions also in master, the more we extend the framework, advanced features, add dodgy xpath queries... the more we are exposed to 3rd party libraries bugs, so for me seems a good option to hardcode them. However, if we do it, we should, at some point, update to the last dependencies minor versions, maybe in the soft freeze to avoid last minute problems.

Thanks for the review David,

Sorry I didn't get which components are you mentioning here David.

My point was to hard-code to that latest version and keep master up-to-date. But when I talked to Damyon, he made a very good point about updating it similar to tinymce or any other 3rd party library. Which is 1 month before release.

So 2.7 and master will follow till last point.

FYI: Current branch testing needs version 1.25.13, 1.26.9 and 1.27.10

Aja, yep, as talked, adding more exact versions, specially in stable is perfect. And yes, we should keep the old (before the crawler fail) version until new releases happen.

So it looks perfect... giving them a try...

Aja, just adding here the way I had to modify my composer.json to be able to get your branches, coz it was really tricky (see the double "dev-"). Sure there is another way but I was not able to get it:

{

    "repositories" : [

        {

            "type": "vcs",

            "url": "https://github.com/rajeshtaneja/moodle-behat-extension"

        }

    ],

    "require-dev": {

        "phpunit/phpunit": "3.7.*",

        "phpunit/dbUnit": "1.2.*",

        "moodlehq/behat-extension": "dev-dev-45701-m26"

    }

}

Performing a 26_STABLE run now...

25_STABLE results using dev-45701-m25:

Moodle 2.5.6+ (Build: 20140522), pgsql, a267e328486bab57eda672d10505ae4de9783049

Server OS "Darwin", Browser: "firefox"

271 escenarios (271 exitosos)

9252 pasos (9252 exitosos)

193m29.486s

26_STABLE results using dev-45701-m26:

Moodle 2.6.3+ (Build: 20140522), pgsql, 56f23662b129d6e3fb9493a9eebcfb57b5b1d565

Server OS "Darwin", Browser: "firefox"

356 escenarios (353 exitosos, 3 fallidos)

12308 pasos (12272 exitosos, 33 omitidos, 3 fallidos)

179m22.153s

(the 3 failures are PHP Notices: Undefined property: stdClass::$aliases in /Users/stronk7/git_moodle/testing/mod/glossary/edit.php on line 94, will look/create a issue for them)

27_STABLE results using dev-45701:

Moodle 2.7+ (Build: 20140522), pgsql, 83bc3ccfdf1f3319aa54cc14fe504f05927f20a2

Server OS "Darwin", Browser: "firefox"

01. Xpath matching locator "//li[@id='section-3']/descendant::div[@class='section-modchooser']/span/a" not found.

    In step `Given I add a "Page" to section "3"'. # behat_course::i_add_to_section()

    From scenario `Activity availability display'. # /Users/stronk7/git_moodle/testing/availability/tests/behat/display_availability.feature:43

    Of feature `display_availability'.             # /Users/stronk7/git_moodle/testing/availability/tests/behat/display_availability.feature

404 escenarios (403 exitosos, 1 fallidos)

14014 pasos (13994 exitosos, 19 omitidos, 1 fallidos)

223m26.475s

(the failure was not reproducible in a second execution, I suppose it was a random one)

About the 26_STABLE failures.... (--tags '@block_glossary_random')

0) I was unable to reproduce them in the browser.
1) Using "dev-45701-m26" they happen. Constantly.
2) Using "dev-45701" they don't happen. Constantly.

So, it seems that there is some bug in the "dev-45701-m26" combination of versions causing empty aliases not to be passed or so. I'm not going to create a new issue because it seems directly related with this.

Ciao

Strange,

for me 26 is passing

344 scenarios (344 passed)

12056 steps (12056 passed)

189m49.665s

Aha... It's Moodle 2.6.2+ (Build: 20140502), pgsql, f2791d4944e66ed18a5edd13bd6b063527c45310
Server OS "Linux", Browser: "firefox". Seems something added later, checking on 2.6.3+

Aha.. Fixed it.

I should have used 2.3.13 rather than older version.

Fixed it, running test on nightly...

As per master it's passing for me. Probably a random fail there, running whole suite locally to ensure.

Cause of problem was https://github.com/symfony/DomCrawler/commit/06357a33e4b33351a37f5ae5c27259bbd8f96007
Textarea was not returning value in glossary, by setting a default (setDefault), worked ....

Anyway, this was fixed and 2.3.13 passed that failure point. Waiting for the whole suite to finish.

Hack works, https://github.com/rajeshtaneja/moodle/compare/moodle:master...wip-mdl-45698 for now
Not sure why email gets reset randomly, because of which form is not saved and next step fails. Above patch, removes that randomness.

Aha,

26_STABLE with dev-45701-m26 applied. All ok.

Moodle 2.6.3+ (Build: 20140522), pgsql, 56f23662b129d6e3fb9493a9eebcfb57b5b1d565

Server OS "Darwin", Browser: "firefox"

356 escenarios (356 exitosos)

12311 pasos (12311 exitosos)

179m29.682s

27_STABLE with dev-45701 and wip-mdl-45698 (e479a07) applied. Ongoing...

Moodle 2.7+ (Build: 20140522), pgsql, c2b31904ec89bcb3756b23bcb83cb03862b2b393

Server OS "Darwin", Browser: "firefox"

404 escenarios (404 exitosos)

14020 pasos (14020 exitosos)

222m20.608s

So it seems we are back to all passing, yay. Next steps then, should be:

1) push the changes to behat-extension and tag them
2) move moodle's composer.json to point to those tags
3) happiness

Thanks, Raj!

I have updated mater pointing to 1.28.3 and 27 branch is different and no need to wait for MDL-45408.

If MDL-45408 has to go next week then I will update master commit pointing to 1.27.10

Thanks Eloy, for all the testing on this.

Thanks Raj - this has been integrated now.

Results on local machine.

Moodle 2.8dev (Build: 20140522), pgsql, f65427b6607be376f5c0aa96e54ad76889513838
Server OS "Linux", Browser: "firefox"
Started at 29-05-2014, 11:10

404 scenarios (404 passed)

14020 steps (14020 passed)

285m5.402s

Moodle 2.6.3+ (Build: 20140522), pgsql, 18c31e25b608b5beaecb3e7a5a6b247532693ff0
Server OS "Linux", Browser: "firefox"
Started at 29-05-2014, 11:10

356 scenarios (356 passed)

12311 steps (12311 passed)

242m9.234s

Moodle 2.5.6+ (Build: 20140522), pgsql, 33fdd3fd874edecf6b417dda6c57ff3e9a6129a5
Server OS "Linux", Browser: "firefox"
Started at 29-05-2014, 11:18

271 scenarios (271 passed)

9249 steps (9249 passed)

254m7.263s

On nightly 26 there is 1 fail, but that is random as it was running all 3 behat's together.
25 and master still running on nightly.

Ok, so it seems this is ready to be considered passed (no matter there are some issues with PhantomJS that has been decided to be moved to different issue for next week).

Am I correct? +-1 ?

Ciao

Yes, it's all good.
+1

Passing this, as behat is passing.

I fail in 27 and master phantomjs only, probably another issue and not related to symfony version fix.

Closing this as fixed, many thanks for your effort!

Some people think architecture is
about the genius sketch; I don't.
Great architecture is a collaboration
among a lot of people
over a long period of time.

– Joshua Prince-Ramus