Re: CSS Grid Layout tests suite

Le 2014-10-30 11:55, Manuel Rego Casasnovas a écrit :
> Hi Gérard,
> 
> On 29/10/14 21:17, Gérard Talbot wrote:
>> Okay. I will review your first tests, during initial phase. And it may
>> take me a few days before making a review as I expect to be very busy 
>> in
>> the next few months.
> 
> That would be great! :-)
> We're not in hurry so even if the review pace is slow, that shouldn't 
> be
> a problem.
> We're working on Grid Layout implementation, so we'd do the work to
> create the test suite in parallel, adding tests step by step.
> The good part is that with your help we'll get used to the process, and
> as time passes, we'll be more and more independent.
> 
> As start point I've written a test plan for CSS Grid Layout, and I've
> made a pull-request to import it in the W3C repository:
> https://github.com/w3c/csswg-test/pull/623
> 
> Afterwards, I'll start to create new pull-requests for tests in GitHub
> (unless you prefer to use a different workflow).

Rego,

I'm not familiar with GitHub as of now. I do not think this prevents me 
from reviewing your tests. I am familiar with Shepherd application
( http://test.csswg.org/shepherd/ ) and Mercurial .

> Probably, we could use milestones and issues too (like other specs are
> doing), in order to follow the progress.
> I don't have permissions to create labels and milestones at this point,
> so I'd be grateful if someone can create a milestone "css-grid-1_dev"
> and a label "spec:grid".

Unfortunately, I can not help you with that.

> 
>> It's just a set of notes (more or less organized) I took over time. 
>> When
>> I see or find an issue or problem in someone's test or even in my own
>> tests, I write a note about it.
>> 
>> The general idea of those notes is what to do in tests that help
>> 
>> a) testers,
>> b) reviewers,
>> c) maintenance of tests and
>> d) reduce test linkages (therefore server load),

A bit more explanations on how to reduce test linkages. Ideally, you 
want to reuse as much as possible already-created-and-available 
reference test files. The references test files that are frequently used 
and reused are prefixed with "ref-".

ref-filled-green-100px-square is referenced by ~= 200 tests
ref-if-there-is-no-red is referenced by ~= 300 tests


Eg.
Florian Rivoal created some 33 tests on CSS Conditional Rules Module 
Level 3 (@supports)

http://test.csswg.org/source/css-conditional-3/

http://test.csswg.org/source/css-conditional-3/at-supports-001.html
...
http://test.csswg.org/source/css-conditional-3/at-supports-033.html

with only 2 reference test files for those 33 tests: that is a good 
reuse policy. I added 6 more tests (and even intentionally and 
deliberately reused the same code indentation of Florian's tests) and 
also reused the same reference file (at-supports-001-ref.html).
For an even better reuse of reference test file, we would have to 
mass-edit only slightly those 39 @supports tests to reference the
ref-filled-green-100px-square.xht
reference test file.



>> 
>> and that reduces condition of mistakes, that speeds or eases their
>> tasks. You should pay more attention to 8a- Avoid single cell with
>> single row tables, 8b- How to test inline-block) and 29- Unneeded,
>> unnecessary, extraneous declarations.
> 
> Yeah, I've already read them, and also the other documentation in
> TestTWF webpage. Really interesting notes.
> 
> Thank you very much,
>   Rego


1 additional note:

{

An ideal test is one that has a very narrow corridor of success and very 
large corridor of failure.

Eg.:
http://test.csswg.org/source/css-backgrounds-3/background-size-009.html
at line 22:
I changed 'height: 100px' to 'height: 50px' to "enlarge" the corridor of 
failure.
More info: 
http://test.csswg.org/shepherd/testcase/background-size-009/#comment-2f7643da4c58

Eg.:
http://test.csswg.org/source/css-backgrounds-3/background-size-014.html
at line 24:
I changed 'width: 200px' for 'width: 50px' to "enlarge" the corridor of 
failure.
More info:
http://test.csswg.org/shepherd/testcase/background-size-014/#comment-f2d2f913b910

An ideal test is one that is very demanding, requiring.

An ideal test is one that checks, verifies one and only one single 
aspect (or single statement) of the spec.

}

I have more notes on many aspects of testing: tests rehabilitation, test 
criticisms, color names and recognition, etc.. and even on how 
psychology of perception affect test results.

Gérard
-- 
Test Format Guidelines
http://testthewebforward.org/docs/test-format-guidelines.html

Test Style Guidelines
http://testthewebforward.org/docs/test-style-guidelines.html

Test Templates
http://testthewebforward.org/docs/test-templates.html

CSS Naming Guidelines
http://testthewebforward.org/docs/css-naming.html

Test Review Checklist
http://testthewebforward.org/docs/review-checklist.html

CSS Metadata
http://testthewebforward.org/docs/css-metadata.html

Received on Thursday, 30 October 2014 20:28:04 UTC