Disambiguate attribute declarations from content attributes #10756
Conversation
The change updates the spec to use the term “attribute declaration” in all cases where what the spec is referring to is actually the syntax for declaring attributes in markup. Otherwise, without this change, the spec uses the same term “attribute” to refer both to attribute declarations in markup and to actual content attributes as they exist in the DOM. And we have evidence of authors being confused due to that ambiguous use of the same term to refer to different things — and evidence suggesting the ambiguous usage promotes the wrong mental model of HTML for authors.
There was a problem hiding this comment.
Hmm. I'm +0 on this. It seems vaguely nice to have this kind of distinction, similar to the distinction between tag and element. But unlike tag/element, we don't have a long history of such a distinction to draw on, and the names "attribute" vs. "attribute declaration" are not as distinct in a helpful way. So it might also be OK to leave things ambiguous.
For example, we currently leave things ambiguous between "attribute values" as seen in markup, and "attribute values" as seen in the DOM. Similarly for "CDATA sections" and "comments".
One thing I worry about with this change is that we may not have carefully audited all usage of "attributes". For example, the following quotes don't seem updated, and maybe they should be:
- The "get an attribute" algorithm
- "When an end tag token is emitted with attributes" (and various other tokenizer references)
- "For example, the parsing of certain named character references in attributes"
- "If the attribute is present, its value must either be the empty string or" (???)
Anyway, as long as we make the suggested fixes to preserve IDs here, and other editors or community members don't have strong arguments against it, I'd be up for merging this.
| data-x="syntax-attribute-value">value</span>, separated by an "<code data-x="">=</code>" character. | ||
| The attribute value can remain <a href="#unquoted">unquoted</a> if it doesn't contain <span>ASCII | ||
| whitespace</span> or any of <code data-x="">"</code> <code data-x="">'</code> <code | ||
| <p><span data-x="syntax-attribute-declarations">Attribute declarations</span> are placed inside |
There was a problem hiding this comment.
We should preserve the old ID by changing the <p> to <p id="syntax-attributes">
| @@ -122938,10 +122939,10 @@ dictionary <dfn dictionary>StorageEventInit</dfn> : <span>EventInit</span> { | |||
| </ol> | |||
|
|
|||
|
|
|||
| <h5>Attributes</h5> | |||
| <h5>Attribute declarations</h5> | |||
There was a problem hiding this comment.
Similarly we should preserve the ID here
| @@ -124064,7 +124068,7 @@ dictionary <dfn dictionary>StorageEventInit</dfn> : <span>EventInit</span> { | |||
| <tr> | |||
| <td><dfn data-x="parse-error-missing-whitespace-between-attributes">missing-whitespace-between-attributes</dfn> | |||
| <td><p>This error occurs if the parser encounters <span | |||
| data-x="syntax-attributes">attributes</span> that are not separated by <span>ASCII | |||
| data-x="syntax-attribute-declarations">attributes</span> that are not separated by <span>ASCII | |||
There was a problem hiding this comment.
Update the text here too, maybe?
The change updates the spec to use the term
“attribute declaration”in all cases where what the spec is referring to is actually the syntax for declaring attributes in markup.Otherwise, without this change, the spec uses the same term
“attribute”to refer both to attribute declarations in markup and to actual content attributes as they exist in the DOM.And we have evidence of authors being confused due to that ambiguous use of the same term to refer to different things — and evidence suggesting the ambiguous usage promotes the wrong mental model of HTML for authors.
/forms.html ( diff )
/index.html ( diff )
/indices.html ( diff )
/introduction.html ( diff )
/parsing.html ( diff )
/syntax.html ( diff )