@@ -485,8 +485,6 @@ Proprietary and Non-standardized Features</h4>
485
485
<h4 id="de-facto">
486
486
Market Pressure and De Facto Standards</h4>
487
487
488
- <p class="feedback"> Added "should not" against encumbered features, per dbaron's request.
489
-
490
488
If a feature is <a>unstable</a> (i.e. the spec has not stabilized yet), yet
491
489
492
490
* at least three UAs implement the feature
@@ -510,56 +508,55 @@ Market Pressure and De Facto Standards</h4>
510
508
pretending it's still "experimental" doesn't help anyone.
511
509
Allowing others to ship unprefixed recognizes that the feature is now de facto standardized
512
510
and encourages authors to write cross-platform code.
511
+ </details>
513
512
514
- <p > Note, though, that the CSSWG must still be consulted to ensure coordination across vendors
513
+ <p class="note" > Note that the CSSWG must still be consulted to ensure coordination across vendors
515
514
and to ensure sanity review by the CSS experts from each vendor.
516
515
Note also that <a>rough interoperability</a> still usually means
517
516
painful lack of interop in edge (or not-so-edge) cases,
518
517
particularly because details have not been ironed out through the standards review process.
519
- </details>
520
518
521
- <p> When exposing such an <a>unstable</a> feature to the Web in a production release,
522
- implementations should support both <a>vendor-prefixed</a> and unprefixed syntaxes
519
+ <h5 id="unstable-syntax" class="no-toc">
520
+ Vendor-prefixing Unstable Features</h5>
521
+
522
+ <p> When exposing such a standards-track <a>unstable</a> feature to the Web in a production release,
523
+ implementations should support <em> both</em> <a>vendor-prefixed</a> and unprefixed syntaxes
523
524
for the feature.
524
525
Once the feature has stabilized and the implementation is updated to match interoperable behavior,
525
526
support for the <a>vendor-prefixed</a> syntax should be removed.
526
527
527
- Note: Anyone promoting <a>unstable</a> features to authors
528
- is urged to describe them
529
- through their standard unprefixed syntax,
530
- and to avoid encouraging the use of the <a>vendor-prefixed</a> syntax
531
- for any purpose other than working around implementation differences.
532
-
533
528
<details class=why>
534
529
<summary> Why?</summary>
535
530
This is recommended so that authors can use the unprefixed syntax to target all implementations,
536
531
but when necessary, can target specific implementations
537
532
to work around incompatibilities among implementations
538
533
as they get ironed out through the standards/bugfixing process.
539
534
540
- Some authors unfamiliar with this approach
541
- and accustomed to the earlier practice
542
- of shipping unstable features only under a prefix
543
- may fail to notice that using the unprefixed syntax
544
- is sufficient to target all implementations,
545
- and may therethore include the prefixed syntax
546
- of the various vendors
547
- followed by the unprefixed syntax.
548
-
549
- Even in the face of this possible misuse,
550
- this approach still beneficial.
551
535
The lack of a phase
552
536
where only the prefixed syntax is supported
553
537
greatly reduces the risk of stylesheets
554
- targetting only this prefixed syntax being written .
538
+ being written with only the vendor- prefixed syntax.
555
539
This in turn allows UA vendors to retire
556
540
their prefixed syntax once the feature is stable,
557
- without risking breaking existing content.
541
+ with a lower risk of breaking existing content.
558
542
It also reduces the need occasionally felt by by some vendors
559
543
to support a feature with the prefix of another vendor,
560
544
due to content depending on that syntax.
561
545
</details>
562
546
547
+ <p class="feedback"> Added "should not" against advocating prefixed syntax,
548
+ per Florian's clarification of dual-syntax usage.
549
+
550
+ Anyone promoting <a>unstable</a> features to authors
551
+ should document them using their standard unprefixed syntax,
552
+ and avoid encouraging the use of the <a>vendor-prefixed</a> syntax
553
+ for any purpose other than working around implementation differences.
554
+
555
+ <h5 id="open-technology" class="no-toc">
556
+ Preserving the Openness of CSS</h5>
557
+
558
+ <p class="feedback"> Added "should not" against encumbered features, per dbaron's request.
559
+
563
560
In order to preserve the open nature of CSS as a technology,
564
561
vendors should make it possible for other implementors
565
562
to freely implement any features that they do ship.
@@ -568,7 +565,6 @@ Market Pressure and De Facto Standards</h4>
568
565
and avoid other obstacles (e.g., platform dependency, licensing restrictions)
569
566
to their competitors shipping the feature.
570
567
571
-
572
568
<h3 id="testing">Implementations of CR-level Features</h3>
573
569
574
570
Once a specification reaches the Candidate Recommendation stage,
0 commit comments