- From: Geoffrey Sneddon <me@gsnedders.com>
- Date: Wed, 18 Jan 2017 17:07:15 +0000
- To: Koji Ishii <kojiishi@gmail.com>
- Cc: www-style list <www-style@w3.org>, "public-css-testsuite@w3.org" <public-css-testsuite@w3.org>, GĂ©rard Talbot <css21testsuite@gtalbot.org>
On Mon, Jan 16, 2017 at 7:19 AM, Koji Ishii <kojiishi@gmail.com> wrote: > 2017-01-13 11:03 GMT+09:00 Geoffrey Sneddon <me@gsnedders.com>: >> >> In principle, the tests fail in all UAs > > > I don't think this is true; Blink imports the tests once every a week or > two, Blink has some anti-alias issue on Mac but not on other platforms. > > Each browsers handle fractional pixels and anti-alias differently, and > you're right that we don't have as much experiences on building stable tests > in vertical writing modes yet as we have in horizontal writing modes. I > can't agree more for us to improve that, but as long as we have enough > browsers passing in automated/manual ways, I do not see this as a spec > issue. When all browsers pass *modified versions of the tests* (because, by definition, the reftests in wpt and csswg-test require pixel-for-pixel identity), I don't think we should consider the testsuite good. After all, if it were good, nobody would feel the need to modify it! Given spec advancement is tied to the testsuite (insofar as it is used to demonstrate interoperability), I think that this should be considered an issue. Yes, it may turn out there's no way to use reftests for as many of the writing modes tests, but we should reach a point where the tests don't need to be modified to be at all useful. /g
Received on Wednesday, 18 January 2017 17:07:48 UTC