Should nullable optional dictionary argument have a default value?

[saschanaz]

The Media Session spec has the following IDL:

dictionary MediaPositionState {
  double duration;
  double playbackRate;
  double position;
};

[Exposed=Window]
interface MediaSession {
  // -- snip --
  void setPositionState(optional MediaPositionState? state);
};

Currently the prose says:

If the type of an argument is a dictionary type or a union type that has a dictionary type as one of its flattened member types, and that dictionary type and its ancestors have no required members, and the argument is either the final argument or is followed only by optional arguments, then the argument must be specified as optional and have a default value provided.

Per my understanding the spec requires a default value for optional nullable dictionaries, but is it intended?

[annevk]

Why is this dictionary made nullable? (It's not clear to me we should allow that at all.)

[saschanaz]

AFAIK somehow an author decided to do that without any real benefit, per my reading.

[annevk]

Filed w3c/mediasession#231. Perhaps we should look into forbidding this? At least when there's no required members? And even then it seems like a poor pattern, but I vaguely recall it being too late for nested dictionaries...

[Ms2ger]

I don't think the prose you quote actually requires a default here (it's not a dictionary type or a union type), but agree that the API doesn't make sense as specced.

[bzbarsky]

void setPositionState(optional MediaPositionState? state);

This is not valid IDL. This is mentioned in the note under https://heycam.github.io/webidl/#idl-nullable-type and in the following normative text under https://heycam.github.io/webidl/#idl-operations>:

If the Type of an operation argument is an identifier followed by ?, then the identifier must identify an interface, enumeration, callback function, callback interface, or typedef.

which explicitly excludes "dictionary" from the list, and the following paragraph, which says:

If the operation argument type, after resolving typedefs, is a nullable type, its inner type must not be a dictionary type.

[annevk]

Thanks, then we can close this I think as there's already a downstream issue.

[saschanaz]

👍 Thanks for the details!

[bzbarsky]

I should note that the IDL parser in Firefox already catches this problem with the IDL, though our error reporting there could be better. I'll file a bug on that later today when I can log into bugzilla again.

@foolip This is another instance of Blink's binding generator not enforcing Web IDL constraints and that causing Google engineers to write invalid spec IDL... I didn't see an existing issue on this, so filed https://bugs.chromium.org/p/chromium/issues/detail?id=997708

[saschanaz]

Note that I found this when I run my webidl2.js validator (which currently incorrectly suggests adding a default value for an invalid nullable dictionary argument) against WPT, maybe we can still catch this type of problems there.

[bzbarsky]

Filed https://bugzilla.mozilla.org/show_bug.cgi?id=1577016 on the Gecko side.