Add the "as" attribute for the <link> element #1449

sideshowbarker · 2016-06-21T15:41:10Z

No description provided.

sideshowbarker · 2016-06-21T15:43:54Z

This at least needs technical review from @yoavweiss before merging.

domenic · 2016-06-21T15:44:34Z

source

+     <th> <code data-x="">as</code>
+     <td> <code data-x="attr-link-as">as</code>
+     <td> <span data-x="concept-request-destination">Destination</span> for a preloaded response (for <code data-x="attr-link-rel">rel</code>="<code data-x="rel-preload">preload</code>")
+     <td> <span data-x="concept-request-destination">destination</span>


I think this column is initial-caps. Maybe "Request destination"?

domenic · 2016-06-21T15:57:36Z

source

@@ -116118,7 +116119,7 @@ interface <dfn>External</dfn> {
    <tr>
     <th> <code data-x="">as</code>
     <td> <code data-x="attr-link-as">as</code>
-     <td> <span data-x="concept-request-destination">Destination</span> for a preloaded response (for <code data-x="attr-link-rel">rel</code>="<code data-x="rel-preload">preload</code>")
+     <td> <span data-x="concept-request-destination">Request destination</span> for a preloaded response (for <code data-x="attr-link-rel">rel</code>="<code data-x="rel-preload">preload</code>")


I meant the third column :P

Err, last column.

yoavweiss · 2016-06-23T16:28:28Z

Also, thanks @sideshowbarker for your work on this! :)

sideshowbarker · 2016-06-27T12:13:19Z

Should it be an enumerated attribute?

@domenic @yoavweiss @igrigorik please see b11ef97 and lemme know if I got it right or not, as far as what we need specified here

zcorpan · 2016-06-27T12:13:53Z

source

@@ -12439,12 +12439,27 @@ interface <dfn>HTMLBaseElement</dfn> : <span>HTMLElement</span> {
   <dd>Also, the <code data-x="attr-link-title">title</code> attribute <span data-x="attr-link-title">has special semantics</span> on this element.</dd>
   <dt><span data-x="concept-element-dom">DOM interface</span>:</dt><!--TOPIC:DOM APIs-->
   <dd>
-<pre class="idl">[<span>HTMLConstructor</span>]
+<pre class="idl">enum <dfn data-x="dom-Destination">Destination</dfn> {<!--


HTML enumerated attributes typically don't use IDL enum, but are DOMString.

HTML enumerated attributes typically don't use IDL enum, but are DOMString.

OK but please see @domenic’s comments at #1449 (comment) and #1449 (comment) and see also, e.g., https://html.spec.whatwg.org/multipage/embedded-content.html#texttrackkind

https://html.spec.whatwg.org/multipage/embedded-content.html#htmltrackelement is the track element's interface and kind there is a DOMString, "limited to only known values". The TextTrack interface's kind does not use "limited to only known values".

https://html.spec.whatwg.org/multipage/infrastructure.html#limited-to-only-known-values is only defined for DOMString types.

HTML enumerated attributes typically don't use IDL enum, but are DOMString.

https://html.spec.whatwg.org/multipage/infrastructure.html#limited-to-only-known-values is only defined for DOMString types.

So I should drop the enum from the IDL and revert to the as IDL attribute just being DOMString.

If so what should I add/change in the IDL to address @domenic’s #1449 (comment) and #1449 (comment) comments?

Yes.

No need to change anything else in the IDL; DOMString + limited to only known values gives us the feature-detection we need, same as input type etc.

Though attributes that are "enumerated attributes" have prose saying how each keyword maps to a state.

I can try to fix this myself with a new commit. :-)

This is another level of indirection. :-)

f6c53bc

@annevk says he doesn't like case-insensitive. I suppose it would be possible to use IDL enum instead.

Any others have opinions about the case-sensitivity of this attribute?

Having HTML duplicate the list of fetch destinations is sucky. Can we have Fetch define this enum? I'm surprised it doesn't already; it seems strange you can set the destiniation only with <link rel="preload"> and not using the fetch API.

Oh hey, Fetch does define the enum for us! https://fetch.spec.whatwg.org/#requestdestination Just use that!!

Oh hey, Fetch does define the enum for us! https://fetch.spec.whatwg.org/#requestdestination Just use that!!

Thanks for catching that. See 772f6ba. I’m very happy RequestDestination is defined in Fetch for us but sad that I didn’t read the IDL in the Fetch spec carefully enough myself to realize it already had it…

zcorpan · 2016-06-27T15:22:25Z

Pls see #1449 (comment) (this comment thread is not-so-usefully hidden from view...)

annevk · 2016-06-27T16:45:53Z

FWIW, I don't know if it currently is, but I made the case on IRC that this should be case-sensitive. Otherwise we limit the namespace of a value defined by other standards. I think generally we should just go for case-sensitive where we can. HTML has been inconsistent on this thus far, case-insensitive only really makes sense for ASCII, and case-sensitive is just simpler to test and implement.

zcorpan · 2016-06-28T11:27:52Z

OK in 143fc37 I changed to IDL enumeration which makes it case-sensitive.

sideshowbarker · 2016-06-28T15:15:42Z

OK in 143fc37 I changed to IDL enumeration which makes it case-sensitive.

LGTM

zcorpan · 2016-06-28T15:40:31Z

Hmm @domenic's comment

Should add a conformance requirement that it be a valid destination.

is not yet addressed.

domenic · 2016-06-28T16:25:56Z

It's missing from the element index list of attributes

domenic · 2016-06-28T16:26:36Z

source

@@ -12491,6 +12517,21 @@ interface <dfn>HTMLLinkElement</dfn> : <span>HTMLElement</span> {
  tokens</span> must only include the tokens from this list that the user agent implements the
  processing model for.</p>

+  <p>The <dfn><code data-x="attr-link-as">as</code></dfn> attribute specifies the <span
+  data-x="concept-request-destination">destination</span> for a preloaded response from the resource


s/preloaded response/preload request

domenic · 2016-06-28T16:28:05Z

source

+     <th> <code data-x="">as</code>
+     <td> <code data-x="attr-link-as">link</code>
+     <td> <span data-x="concept-request-destination">Destination</span> for a preloaded response (for <code data-x="attr-link-rel">rel</code>="<code data-x="rel-preload">preload</code>")
+     <td> "<code data-x="">document</code>"; "<code data-x="">embed</code>"; "<code data-x="">font</code>"; "<code data-x="">image</code>"; "<code data-x="">manifest</code>"; "<code data-x="">media</code>"; "<code data-x="">object</code>"; "<code data-x="">report</code>"; "<code data-x="">script</code>"; "<code data-x="">serviceworker</code>"; "<code data-x="">sharedworker</code>"; "<code data-x="">style</code>"; "<code data-x="">worker"</code>; "<code data-x="">xslt</code>"; or the empty string


How about just Destination?

domenic · 2016-06-28T16:32:11Z

source

@@ -12491,6 +12517,21 @@ interface <dfn>HTMLLinkElement</dfn> : <span>HTMLElement</span> {
  tokens</span> must only include the tokens from this list that the user agent implements the
  processing model for.</p>

+  <p>The <dfn><code data-x="attr-link-as">as</code></dfn> attribute specifies the <span
+  data-x="concept-request-destination">destination</span> for a preloaded response from the resource
+  given by the <code data-x="attr-link-href">href</code> attribute. The possible values are the


Again, duplicating the values is not good IMO. "The possible values are any destination" makes sense to me.

sideshowbarker · 2016-06-29T02:45:06Z

It's missing from the element index list of attributes

Added in f4576b1

sideshowbarker · 2016-06-29T02:54:34Z

Hmm @domenic's comment

Should add a conformance requirement that it be a valid destination.

is not yet addressed.

...

Again, duplicating the values is not good IMO. "The possible values are any destination" makes sense to me.

I think for authors it’s ideally better for the allowed values to be explicitly listed at point of use in the HTML spec itself, rather than requiring them to also check another spec just to determine what the document-conformance requirements are (especially in the case like this where the microsyntax of the attribute is really simple—just one keyword among an enumerated list of keywords).

But that said, duplicating values is also not good and should be avoided.

So I’m fine personally in this case with just going back to having this change to the HTML spec state that the value must be a valid destination, with a link to the definition in the Preload spec. But would like to wait to find out if @zcorpan agrees.

zcorpan · 2016-06-29T09:19:07Z

I suppose the normative part of the spec should link to fetch, and the index can do that also (input type in the index already has a link instead of enumerating all the values), but it would be nice with some non-normative text inline saying what they keywords are.

sideshowbarker · 2016-06-29T11:23:09Z

LGTM after daa16c6

I suppose the normative part of the spec should link to fetch, and the index can do that also (input type in the index already has a link instead of enumerating all the values)

Yup, as I see you made it.

but it would be nice with some non-normative text inline saying what they keywords are.

Yup, just putting it into a Note as you made it seems sufficient.

domenic · 2016-06-29T16:22:08Z

I'm really not a fan of any duplication. It just means more maintenance whenever Fetch updates. This has already been a pain for referrerpolicy that is causing HTML and Referrer Policy to get out of sync (see #1419). I think it's OK to refer authors to another spec to see the valid values.

domenic · 2016-06-29T16:23:01Z

source

+  if the corresponding content attribute's value <span
+  data-x="case-sensitive">case-sensitively</span> matches one of the enumerated values, then the IDL
+  attribute must return the content attribute's value; otherwise it must return the content
+  attribute's default value; and on setting, the content attribute must be set to the specified new


Maybe s/; and on/. On/

domenic · 2016-06-30T19:21:05Z

Lovely work everyone! Thanks @yoavweiss for being willing to integrate like this. I think authors and tools really benefit from having all the relevant attributes of an element in one place.

yoavweiss · 2016-07-01T06:40:25Z

And thanks @sideshowbarker and @zcorpan for actually working on this while I was traveling! :D

Add the "as" attribute for the <link> element

5578643

sideshowbarker mentioned this pull request Jun 21, 2016

Add preload link type #1418

Closed

sideshowbarker assigned yoavweiss Jun 21, 2016

domenic reviewed Jun 21, 2016
View reviewed changes

domenic added the addition/proposal New features or enhancements label Jun 21, 2016

fixup! Add the "as" attribute for the <link> element

94e9105

domenic reviewed Jun 21, 2016
View reviewed changes

sideshowbarker added 2 commits June 22, 2016 01:01

fixup! Add the "as" attribute for the <link> element

d5db2d9

fixup! Add the "as" attribute for the <link> element

ae60228

fixup! Add the "as" attribute for the <link> element

b11ef97

sideshowbarker unassigned yoavweiss Jun 27, 2016

zcorpan reviewed Jun 27, 2016
View reviewed changes

Don't use IDL enum; specify invalid/missing value defaults

f6c53bc

zcorpan added the do not merge yet Pull request must not be merged per rationale in comment label Jun 27, 2016

Switch to IDL enumeration to make the value case-sensitive

143fc37

zcorpan removed the do not merge yet Pull request must not be merged per rationale in comment label Jun 28, 2016

zcorpan added the do not merge yet Pull request must not be merged per rationale in comment label Jun 28, 2016

domenic reviewed Jun 28, 2016
View reviewed changes

Add document conformance requirement about as's value

b1fbd44

domenic reviewed Jun 28, 2016
View reviewed changes

sideshowbarker added 2 commits June 29, 2016 11:38

Add “as” to element index list of attributes

f4576b1

Change “preloaded response” to “preload request”

084df38

Reference RequestDestination for Fetch

772f6ba

Don't normatively duplicate the destination keywords

daa16c6

zcorpan removed the do not merge yet Pull request must not be merged per rationale in comment label Jun 29, 2016

zcorpan assigned domenic Jun 29, 2016

domenic reviewed Jun 29, 2016
View reviewed changes

Address @domenic's comments (remove duplication; grammar)

935e8d3

domenic merged commit 2381476 into master Jun 30, 2016

domenic deleted the as-attribute branch June 30, 2016 19:20

This was referenced Aug 11, 2016

Consider not repeating the referrer policy values so many times #1656

Closed

Case-sensitive as="" for <link> is weird #1665

Closed

domenic added the topic: link label Jun 15, 2021

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add the "as" attribute for the <link> element #1449

Add the "as" attribute for the <link> element #1449

sideshowbarker commented Jun 21, 2016

sideshowbarker commented Jun 21, 2016

domenic Jun 21, 2016

domenic Jun 21, 2016

domenic Jun 21, 2016

yoavweiss commented Jun 23, 2016

sideshowbarker commented Jun 27, 2016

zcorpan Jun 27, 2016

sideshowbarker Jun 27, 2016

zcorpan Jun 27, 2016

sideshowbarker Jun 27, 2016

zcorpan Jun 27, 2016

zcorpan Jun 27, 2016

zcorpan Jun 27, 2016

domenic Jun 28, 2016

domenic Jun 28, 2016

sideshowbarker Jun 29, 2016

zcorpan commented Jun 27, 2016

annevk commented Jun 27, 2016

zcorpan commented Jun 28, 2016

sideshowbarker commented Jun 28, 2016

zcorpan commented Jun 28, 2016

domenic commented Jun 28, 2016

domenic Jun 28, 2016

domenic Jun 28, 2016

domenic Jun 28, 2016

sideshowbarker commented Jun 29, 2016

sideshowbarker commented Jun 29, 2016

zcorpan commented Jun 29, 2016

sideshowbarker commented Jun 29, 2016

domenic commented Jun 29, 2016

domenic Jun 29, 2016

domenic commented Jun 30, 2016

yoavweiss commented Jul 1, 2016

Add the "as" attribute for the <link> element #1449

Add the "as" attribute for the <link> element #1449

Conversation

sideshowbarker commented Jun 21, 2016

sideshowbarker commented Jun 21, 2016

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

yoavweiss commented Jun 23, 2016

sideshowbarker commented Jun 27, 2016

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

zcorpan commented Jun 27, 2016

annevk commented Jun 27, 2016

zcorpan commented Jun 28, 2016

sideshowbarker commented Jun 28, 2016

zcorpan commented Jun 28, 2016

domenic commented Jun 28, 2016

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

sideshowbarker commented Jun 29, 2016

sideshowbarker commented Jun 29, 2016

zcorpan commented Jun 29, 2016

sideshowbarker commented Jun 29, 2016

domenic commented Jun 29, 2016

Choose a reason for hiding this comment

domenic commented Jun 30, 2016

yoavweiss commented Jul 1, 2016