We could cut down on CSS error spam by not reporting errors for non-Gecko, vendor-specific identifiers which follow CSS2.1 rules. Consequently, I propose we not report errors for any identifier meeting the following criteria: - begins with '-'/'_' - but not with '-moz-', to enable typo-catching The only down side I see is if someone mistypes the -moz- as -mzo- or similar (and won't get an error). I expect this occurs much less often than using other-vendor or IE-workaround identifiers, and I'd bet you're less likely to misread and mistype '-moz-' than most strings. The following patch removes these errors for property names and at-rules. Note that nsICSSParser::ParseProperty doesn't need to be changed because it can only be called with an nsCSSProperty (and the only unhandled one is eCSSProperty_UNKNOWN). I've used the following URL as a testcase (should only get warnings for @-moz-e and -moz-e): data:text/html,<style>@-moz-e{}@-khtml-f{}f{-moz-e:foo;-khtml-e:bar;_width:2px;-width:3px;}</style> At a glance, doing this for other parts of CSS requires decentralized fixes or parser cleanup, so I'm leaving property values, pseudo-selectors, etc. for future patches.

This needs review from dbaron

If we do this then we should probably ignore mso- as well. http://www.w3.org/TR/CSS21/syndata.html#vendor-keyword-history

Comment on attachment 278288 [details] [diff] [review] Ignore other-vendor @at-rules and property names Okay; I figured I'd try to avoid dbaron since he's out-of-townish for awhile. He was okay with the general idea when I mentioned it to him a few weeks ago, although I mentioned doing it only for identifiers matching /^-.+-/ but not /^-moz-/ because I hadn't read that section of CSS2.1 before. As for mso-, I think we should only ignore what the spec says can be ignored. I doubt many not-saved-Word-docs use it, and this is mostly to help web developers.

Comment on attachment 278288 [details] [diff] [review] Ignore other-vendor @at-rules and property names >+ return (ident.First() == PRUnichar('-') || ident.First() == PRUnichar('_')) && >+ !StringBeginsWith(ident, NS_LITERAL_STRING("-moz-")); You could put the "&& !StringBeginsWith()" inside the first half of the ||. r+sr=dbaron Sorry for the long delay.

Cuts down on error console spam for site authors that use IE hacks or include code tailored to the CSS support of other browsers...

Comment on attachment 283496 [details] [diff] [review] Address comment Checked in; I'll get to the rest of the things eventually, although not likely soon given that the checks are currently spread throughout the parser and will either require a careful hand or (more likely) refactoring to make fewer checks necessary.

Comment on attachment 283496 [details] [diff] [review] Address comment Resetting all approval1.9+ flags on bugs that have not been checked in by Oct 22 11:59 PM PDT. Please re-request approval if needed.

Comment on attachment 283496 [details] [diff] [review] Address comment Landed by Waldo on 2007-10-03 21:58.

Comment on attachment 283496 [details] [diff] [review] Address comment Reset approval flag to + as it was already checked in.

Resolving per comment 9

I filed bug 433074 as followup to fix the remaining issues, although I'm not 100% sure what they all are.