W3C home > Mailing lists > Public > public-html-testsuite@w3.org > April 2013

some WebIDL testing / documentation

From: Dzenana Trenutak <dzenana.trenutak@gmail.com>
Date: Thu, 4 Apr 2013 13:31:29 -0400
Message-ID: <CAPDCRJTNXad4c7v3JUCrLBqMMD7Wn=6pON_txBY0c+nN_oTHWA@mail.gmail.com>
To: public-test-infra-request@w3.org, public-html-testsuite@w3.org, Robin Berjon <robin@berjon.com>, Tobie Langel <tobie@w3.org>

OK, I've added what I gleaned about writing WebIDL tests up
updating all around to match the new repo name while I was at it.  (Tobie,
you saw that I signed the CLA back in March?  Let me know if the new name
-> new link requires a new CLA too.)

And I just submitted a pull request providing the WebIDL tests meant to
cover the HTML "sections" and "grouping-content" sections.  However, you'll
see that there's not just 1 but 5 flavors of each test file, as I'm not
sure which approach is the most desired:

- the "dummies" approach which I saw in existing tests (the "untested" IDL
is a bare skeleton)  (-->  ~50 tests / file)

- the "full IDL" approach, in which the "untested" IDL contains all the
background WebIDL to support the "tested" IDL  (-->  ~1500 tests / file)

- the "all tested" approach, in which *all* the IDL is tested along with
the object(s), because, as was recently written on the list, chugging
through many / redundant tests is cheap, so "why not".  (-->  ~2000 tests /

- the "prevent multiple" approach, by which many interfaces can be just
tested once within each file.  This may be contrary to the "why not"
approach, but the API exists, so here it is being used (although separate
test files still duplicate tests)  (--> ~250 tests / file)

- the "prevent multiple but test all IDL" approach, which is perhaps an odd
compromise but aims to maximize the number of unique tests performed.  (-->
 ~850 tests / file)

I'll be curious to see which one you cherry pick to include in the test
suite (if any), and will be happy to update the documentation accordingly
(right now, the documentation describes the "full IDL" approach).

Thank you in advance for any feedback.  :-)


Dzenana T.
Received on Thursday, 4 April 2013 17:31:58 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:14:32 UTC