Engineer on @googlechrome. Involved in CSS and W3C standards. Previously @mozilla, @w3ctag. Mastodon: @dbaron@w3c.social

Rockville, Maryland, USA
Joined March 2008
Filter
Exclude
Time range
-
Near
Replying to @meyerweb
@meyerweb Cookie confirmation is a massive impedance mismatch between synchronous and asynchronous APIs, and thus totally broken.
1
Replying to @mirceadesign
@mirceadesign We want to remove it as soon as we can.
2
@thierrykoblentz @tabatkins Either not fixed or no transform. Why the choice was not fixed... maybe ask @smfr ?
1
Replying to @paul_irish
@paul_irish @paulrouget I think that's *usually* true for Gecko as well, but some pages hit really bad selector matching cases. cc: @bz_moz
2
Replying to @johnolilly
@johnolilly That's pretty normal for late March through late May (or whenever it actually warms up in the bay area), roughly.
Replying to @meyerweb
@meyerweb (color-index) is false because the device uses RGB values rather than using a lookup table (256-color, 16-color, etc.)
1
Replying to @meyerweb
@meyerweb dev.w3.org/csswg/css3-mediaq… says "(feature) will evaluate to true if (feature:x) will evaluate to true for a value x other than zero ..."
1
Replying to @JamesFallows
@JamesFallows Having waited as long as 90 mins for a checked bag, I aimed for carry-on only before checked bag fees were an issue.
1
How Green Are Electric Cars? It matters how the electricity where you plug in is produced. nyti.ms/IlLhVI
1
Replying to @tabatkins
@tabatkins Or if you store the sign information in the number tokens
Replying to @tabatkins
@tabatkins @elwoz An alternative for :nth-child() is to concatenate the CSS tokens and rescan them.
2
Replying to @elwoz
@elwoz @tabatkins Yeah, :nth-child() is bad. It should maybe be in the tokenizer instead of the parser, but url() didn't turn out so well.
Replying to @tabatkins
@tabatkins It better be. They're separate layers, and they should stay separate layers.
2
Replying to @tabatkins
@tabatkins No, it always tokenizes as CDO, but CDO token is a parse error anywhere but top level. Tokenizer can't depend on parser state.
1
Replying to @tabatkins
@tabatkins That sounds like something for a parsing spec, not for a tokenizing spec.
2
Replying to @tabatkins
@tabatkins But the optional sign before the digit means it requires two character lookahead. 7e+3 is number, 7e+a is 3 tokens, 7e-a is dim
1
1
Replying to @tabatkins
@tabatkins did you included SVG's added scientific notation? (Fun to distinguish from units.)
2
Replying to @tabatkins
@tabatkins No, because you have to match parens for baduri just like you do for ( or function(
1
@johnath while you're in sf... have you seen the grebes in the bay? (I haven't paid enough attention to say they're Clark's vs. Western.)
Replying to @hober
@hober @sgalineau @smfr I don't see any conformance requirements, so no testcases are needed.