Exclude all problematic pseudo-classes from rule-merging #392
Conversation
|
Seems the first commit made it into this PR too because I based the second branch on it. |
| @@ -13,7 +13,7 @@ var DEFAULTS = { | |||
| }, | |||
| selectors: { | |||
| ie7Hack: false, // *+html hack | |||
| special: /\-(moz|ms|o|webkit)\-/ // special selectors which prevent merging | |||
| special: /(\-moz\-|\-ms\-|\-o\-|\-webkit\-|:dir\((ltr|rtl)\)|:first|:fullscreen|:left|:read-only|:read-write|:right)/ // special selectors which prevent merging | |||
There was a problem hiding this comment.
Two questions:
- Shall we rather just check for
:dirin the pattern? - Checking for
:firstmay also match:first-childwhich is mergeable.
There was a problem hiding this comment.
Shall we rather just check for :dir in the pattern?
Yeah, might be more proof to future changes. Something like /:dir\(.+\)/ will probably be better.
Checking for :first may also match :first-child which is mergeable.
Good point. I think I'd use something like /:first(?!-)/ to avoid this, if you have a better suggestion, go ahead. I'm not 100% sure if this is even an issue, or if the class detection is good enough to avoid such a mismatch.
There was a problem hiding this comment.
Maybe these are better. I'm a noob when it comes to negative lookahead. Feel free to improve :)
/:dir\([a-z-]*\)/
/:first(?![a-z-])/Edit: updates
There was a problem hiding this comment.
@silverwind These look good. Could you make a change and also add testcases for :first and :first-child?
|
Sorry for those extra commits on the tests, now it's done :) |
|
Feel free to squash the last 3 together. Otherwise I'll do it :) |
silverwind
force-pushed
the
pseudo-class-merge
branch
from
8674a5b
to
20b0c6c
Compare
|
And they're squashed. |
Exclude all problematic pseudo-classes from rule-merging
|
💯 now off to backporting it into 2.2. |
|
👍 looking forward to it! |
|
|
Here's the patch to exclude all seven problematic pseudo-classes from rule-merging as discussed in #390.