[tor-bugs] #5290 [Quality Assurance and Testing]: Collect+write test pages for JavaScript hooks

[Tor Bug Tracker & Wiki]

#5290: Collect+write test pages for JavaScript hooks
-------------------------------------------+--------------------------------
 Reporter:  mikeperry                      |          Owner:  cypherpunks                  
     Type:  task                           |         Status:  new                          
 Priority:  major                          |      Milestone:  TorBrowserBundle 2.3.x-stable
Component:  Quality Assurance and Testing  |        Version:                               
 Keywords:  MikePerry201208                |         Parent:  #5292                        
   Points:                                 |   Actualpoints:                               
-------------------------------------------+--------------------------------

Comment(by gk):

 Replying to [comment:11 mikeperry]:
 > gk: this is awesome. I am wondering two things: First, what made you
 pick mozmill over mochitest or xpcshell? Or for that matter, anything else
 out of https://developer.mozilla.org/en/Mozilla_automated_testing?
 Well, the overall goal I had in mind was to have just one testing
 framework for everything reachable by JavaScript. That includes add-on
 testing (Torbutton AND Torbirdy) and Torbrowser testing. Add-on testing
 (functionality and UI) is not possible with xpcshell and Mochitest is
 supposed to be a browser-based tool which therefore does not seem fit to
 the Torbirdy requirement (you'll find the browser-based argument and other
 interesting points about mochitest vs. mozmill here:
 https://groups.google.com/group/mozmill-
 dev/browse_thread/thread/c7170970800f6e79/9a4c8721310412ee?hide_quotes=no).
 Looking at the other available tools and the above mentioned requirements
 made me believe as well we should give Mozmill a try. Sure it is no catch-
 all, we may need compiled-code tests but having just to types of tests to
 maintain and develop is much, much easier than half a dozen types. That
 said, I am open for (even) better solutions, though.
 > Second, what do you propose we should do with this zip? Does Firefox
 have Makefile rules for mozmill we should use?
 It does not have those yet. The current tests are in a spearate repo,
 http://hg.mozilla.org/qa/mozmill-tests but that is mainly due to not using
 Mozmill as a unit/whitebox testing tool (which it could be, though) as far
 as I understand it. Therefore, Mozmill itself is not included in the main
 tree (mozilla-central) but must be installed separately to run the tests.

 > If Makefiles are a poor plan, should we create a new testing repo for
 stuff like this?
 I don't think Makefiles are a poor plan. It all depends on how the tests
 shall be used or better how the testing infrastructure should look like in
 the end, a thing that should probably discussed somewhere else with a
 wider audience than this bug as it is off topic and more important than
 the particular tests I attached. That said, if we want to start with some
 sort of blackbox/greybox tests like the ones I inlcuded, an own repo and
 some tweaks to Torbutton (the modal dialog on start-up) should be enough
 to get the tests integrated smoothly into the Torbutton/Torbrowser
 development workflow, I guess.

-- 
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/5290#comment:12>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online