Re: Mismatch between CSS and web-platform-tests semantics for reftests

On 08/19/2014 03:01 AM, James Graham wrote:
> web-platform-tests has a rather simple set of semantics for reftests;
> each test is identified by a single (test, ref, type) tuple. Multiple
> <link> elements in the same test file result in multiple tests with
> different references.
>
> By contrast the testthewebforward documentation suggests that CSS has a
> rather more complex semantic; each test is of the form (test, ref_tree),
> where ref_tree is a tree of (file, type) nodes and breadth in the tree
> indicates multiple options and depth in the tree indicates multiple
> requirements, so the overall test passes if all the comparisons down to
> some leaf node pass. In concrete terms multiple <link> elements in the
> same document represent multiple options, and each reference can <link>
> to further references.
>
> Do we know how necessary the more complex setup is? In particular I
> think that the main distinguishing feature is the ability to specify
> multiple references of which *any* may match (since the all must match
> semantic is just a way of collapsing N tests down to a single test). Is
> there a list of all tests that are using this feature?

I'm not sure about tests using this feature currently, but there are
some branches in the CSS spec that would warrant it. E.g. 'outline'
can be painted at one of two places in the tree. Or letter-spacing
can either perform cursive elongation on Arabic -or- not affect its
letters at all but cannot put gaps between them.

We could remove the feature to match web-platform-tests, but then
how would we handle those cases?

~fantasai

Received on Tuesday, 19 August 2014 15:50:19 UTC