Don't test for prefixed support.

[dbaron]

This patch removes the tests for prefixed support so that only support
for unprefixed implementation of the features is tested.

I realize it's not obvious that this change is the right thing to do,
but I think that given current thoughts on prefixing and its influence
on the Web, I think this change is worth considering.

Widespread browser support for prefixed CSS has led to large amounts of
browser-specific CSS on the Web; promoting support for unprefixed CSS
helps encourage browsers to behave in a way that encourages
cross-browser content on the Web.

[LeaVerou]

Hi David,

Thanks for the effort. However, as you can probably guess, I cannot merge this.
As much as I want browsers to move towards unprefixed CSS, the goal of css3test is to help authors, and authors do not consider prefixed support to be equivalent to no support, so I'd be doing a disservice to them if I didn't check for prefixed support.

That said, perhaps we could penalize prefixed support somehow? Maybe it would be counted as a half pass or it would have a warning or something?

[dstorey]

It is a drawback of the test that you can't tell if it it is supported regularly or needs a prefix. A different colour would be nice (with label). Penalising would be good too. I think it is fair enough, as for browsers that are not webkit, they often have to go back and implement the webkit version (which is not always the same and needs work to be bug and behaviour compatible), which creates extra work and is often at the expense of implementing something else (or fixing more interop bugs in another area)

Also, when features are renamed, the browser is already penalised by losing all points (text-combine-horizontal, which became text-combine-upright for example). There isn't much difference between a prefixed implementation and a renamed implementation.

[LeaVerou]

Yup, totally agreed. Penalizing would be great. Dropping prefix testing entirely, not so much.