Re: border-001-018: updates and conversions

Le Mer 12 septembre 2012 15:02, Rebecca Hauck a écrit :
> Hi Gérard,
> I've updated the first batch of borders tests to include CSS3 spec links
> and convert wherever possible. Since you've been doing a lot of
> conversions, I wanted to have you spot-check this first batch to make
> sure the changes I've made are ok.
> Most/all of the tests in this suite are testing the border shorthand
> property – not specific property values, just various permutations &
> syntax.  I took the liberty of changing dashed borders to double in
> order to easily create references.


Here are some preliminary comments.

"double was better to use for reftests"

As far as I can see, border-style: double is not reftestable in CSS 2.1
and in CSS 3:

    Two parallel solid lines with some space between them. (The
thickness of the lines is not specified, but the sum of the lines
and the space must equal ‘border-width’.)

So, 2x + y == border-width : that's unsolvable as there are many

If border-width is 12px, then how can you be sure that individual lines
are each 4px and in-between-lines space is 4px; why it couldn't be 3px
for individual lines and 6px for the in-between-lines space ?

> Can you confirm that's ok?  The rest
> of the changes I made were smaller – increased width of the border,
> change "blue" to "green" in some cases.

it's best to also avoid green unless using the presence of red to
indicate failures.

> Also, the wording of the assert
> I changed from "two boxes" to "two concentric squares" (for example).
> The latter is less ambiguous.

In my opinion, almost anything which is reasonably descriptive is better
than "box". To a wide majority of people, a box is a 3-dimensional

Sometimes, it's not really important to replace the word "box" in a test:
eg.: the color of 2 shapes is the object of the test

On the other hand, often, to help the tester, it is important to
describe the shape of rendered objects.

The balance between overdoing it (thus making the sentence to read
longer) and underdoing it is not always obvious.

eg.: I need to tweak the pass/fail conditions sentence of this test:

eg.: the pass/fail conditions sentence of this test:
seems overly descriptive

> The tests where I only added spec links I pushed back to the approved
> directory:  border 001, 003, 005, 006, 008. (maybe these can be flipped
> back to Approved now?)
> The ones I converted and changed are forked and pushed to my submitted
> folder: rhauck/submitted/css3-backgroundsborders:  border 002, 004, 007,
> 009-018.
> Also, one last question:  can you clarify the proper tagging I should be
> doing in the subject line for other mails of the nature?  Should I be
> adding [RC6] to them as you do?

I add [RC6] so that eventual searching through archives will be
assisted. So, yes, this helps.

The most important chunk of the email's subject line is the base
filename of the related test.

Contributions to the CSS 2.1 test suite:

CSS 2.1 Test suite RC6, March 23rd 2011:

CSS 2.1 test suite harness:

Contributing to to CSS 2.1 test suite:

Received on Wednesday, 12 September 2012 21:03:05 UTC