[css-mixins-1] @function grammar is missing open-paren

[ferdnyc]

The definition of the @function syntax is:

<pre class="prod def" nohighlight>
&lt;@function> = @function <<function-token>> <<function-parameter>>#? )
	[ returns <<css-type>> ]?
{
	<<declaration-rule-list>>
}

<dfn><<function-parameter>></dfn> = <<custom-property-name>> <<css-type>>? [ : <<default-value>> ]?
<dfn><<css-type>></dfn> = <<syntax-component>> | <<@function/type()>>
<dfn><<default-value>></dfn> = <<declaration-value>>
<dfn function lt="type()" for="@function">&lt;type()></dfn> = type( <<syntax>> )
</pre>

The literal close-paren ()) that follows the <function-parameter> list is represented, but the open paren (() that should fall between the <function-token> and the (optional) parameter list is not.

The simplest solution is for the grammar to read:

<pre class="prod def" nohighlight>
&lt;@function> = @function <<function-token>> ( <<function-parameter>>#? )
	[ returns <<css-type>> ]?
{
	<<declaration-rule-list>>
}

<dfn><<function-parameter>></dfn> = <<custom-property-name>> <<css-type>>? [ : <<default-value>> ]?
<dfn><<css-type>></dfn> = <<syntax-component>> | <<@function/type()>>
<dfn><<default-value>></dfn> = <<declaration-value>>
<dfn function lt="type()" for="@function">&lt;type()></dfn> = type( <<syntax>> )
</pre>

Unless <function-token> is meant to include the open-paren, but since it isn't defined that way anywhere (formally or informally), that seems unlikely.

(Also, it would make more sense structurally if <function-token> were then defined to include not only the open-paren, but also the argument list and the close-paren. Its definition could then be decomposed further.)

[ferdnyc]

@tabatkins reports that <function-token> does include the open-paren, in which case I guess this becomes a report that <function-token> isn't formally defined anywhere.

[tabatkins]

? Sure it is, https://drafts.csswg.org/css-syntax/#typedef-function-token

[astearns]

Well, the open-paren-consuming part is defined in https://drafts.csswg.org/css-syntax/#consume-ident-like-token

[ferdnyc]

Ah, so it is, thanks @astearns.

It is a bit unfortunate that that aspect isn't part of any of the formal grammar productions anywhere, as (for example) MDN imports the @function grammar (and only the grammar) directly from the standards documents, and therefore lacks any definition of <function-token>.

But something can probably be added to the data on their end, to provide a definition.

Closing this as addressed, though, apologies for the noise.

[tabatkins]

The <*-token> definitions are the bottom of the grammar productions; below that there's only parsing algorithms. At some point you have to stop and switch to something imperative.