Add unique ids to each parse error

[gsnedders]

It would be convenient for the sake of testing implementations to have unique ids on each parse-error.

I know @Hixie opposed this because the spec only requires parse errors to emitted at specific points, and some implementations merge multiple parse errors in the spec, and hence cannot distinguish them.

I don't think this is a good argument against it, as implementations can always have a mapping of their parse error x corresponds to a and b in the spec.

At the moment html5lib-tests essentially has position and some meaningless message.

Both html5lib and html5ever have unique messages for each parse error. It would be nice to make sure we have the right errors being raised.

We probably want ids something like the keys in the dictionary E in https://github.com/html5lib/html5lib-python/blob/master/html5lib/constants.py#L7.

[gsnedders]

cc @zcorpan @nox @hsivonen

[zcorpan]

Sounds like a good idea. (I also think the parser could use more ids in general.)

[hsivonen]

I'm OK with this.

[haroldfredshort]

I've gone through the microsyntax parse errors first. There are not very many but these are the hardest to properly identify. With one exception they can all fit the 'x-in-y' pattern. What do you think of these so far?

parse-error-comma-in-img-srcset
parse-error-trailing-commas-in-img-srcset-url
parse-error-invalid-value-in-img-descriptor
parse-error-img-descriptor (this is an else-case parse error for img descriptors, not sure what to call it)
parse-error-empty-size-in-sizes-attribute
parse-error-invalid-size-in-sizes-attribute

Example:
<span data-x="concept-microsyntax-parse-error" id="parse-error-comma-in-img-srcset">parse error</span>

Next I'll see about listing the generalized forms for the remaining parse errors.

[sideshowbarker]

I think we should consider actually uniquely naming each parse error in the body of the spec; like:

<code><dfn>CommaInImgSrcset</dfn></code><span
    data-x="concept-microsyntax-parse-error">parse error</span>

[domenic]

Gah, we keep giving poor @haroldfredshort conflicting information :(...

[domenic]

Maybe people would all agree on something like

... that is a <span data-x="concept-microsyntax-parse-error">parse error</span>
(<dfn data-x="parse-error-comma-in-img-srcset">comma in img srcset</dfn>).

???

[sideshowbarker]

Gah, we keep giving poor @haroldfredshort conflicting information :(...

oofs, yeah, sorry for not re-reading the comments in the PR more carefully

Maybe people would all agree on something like

... that is a <span data-x="concept-microsyntax-parse-error">parse error</span>
(<dfn data-x="parse-error-comma-in-img-srcset">comma in img srcset</dfn>).

Yeah, that would be fine by me.

[zcorpan]

I believe this bug is about parse errors in the "Parsing HTML documents" section -- not about errors in srcset.

[haroldfredshort]

Regarding the comment from @zcorpan, should I not add identifiers to the seven srcset parse errors, or would it be okay to identify these differently? Like this:

that is a <span data-x="concept-microsyntax-parse-error">parse error</span>
(<dfn data-x="microsyntax-parse-error-comma-in-img-srcset">comma in img srcset</dfn>).

The parse errors in the "Parsing HTML documents" section are simpler, but before I go further I am including an example here to make sure we are all agreed that this is how we want it to be:

...
<dt>EOF</dt><dd><span>Parse error</span>
(<dfn data-x="parse-error-eof-in-cdata-section-state">EOF in CDATA section state</dfn>)
...