829248 - Implement support for [TreatUndefinedAs=Missing] in WebIDL

Status: RESOLVED FIXED

(Core :: DOM: Core & HTML, defect)

Description

[Boris Zbarsky [:bzbarsky]]

I want this for add/removeEventListener.

Comment 1

[Boris Zbarsky [:bzbarsky]]

Actually, I don't.  But this is still useful!

Comment 2

[Josh Yuan]

Hello there,

My friend and I are interested in helping out.  Is this bug still active?  If so, where/how do we begin?

Thanks.

Comment 3

[Boris Zbarsky [:bzbarsky]]

This is still active, yes.

Assuming you've pulled a Firefox tree, you want to look at dom/bindings/Codegen.py and read http://dev.w3.org/2006/webapi/WebIDL/#dfn-overload-resolution-algorithm I guess.  Then start thinking about how to best implement it (probably by walking backwards through the arguments and changing argc as neeed up front in the generated code for method calls).

Comment 4

[Josh Yuan]

Cool!

I assume this is part of a large(r) project overall, are there any other mentored bugs from this project that I could work on while we're at it?

Thanks!

Comment 5

[Boris Zbarsky [:bzbarsky]]

You can search bugzilla for mentored bugs, right?

Comment 6

[Zuhao(Joe) Chen]

I'm not familiar with the Implementation of WebIDL but can I take this bug and try to work on it?

Comment 7

[Boris Zbarsky [:bzbarsky]]

Sure.  Please feel free to ask questions as needed!

Comment 8

[Zuhao(Joe) Chen]

I've read the w3c website about WebIDL and I kinda understand it's the interface between JavaScript and browser backend but I'm not sure how the implementation reflects to firefox code base.

Could you briefly explain it to me or point me to a page where I can read about it myself?

Thanks!

Comment 9

[Boris Zbarsky [:bzbarsky]]

The way it works right now is that there is a Python script (well, several) that parses WebIDL and produces C++ code that acts like glue between JavaScript and the C++.

The relevant code is in dom/bindings/parser/WebIDL.py (the parser) and dom/bindings/Codegen.py (the script that builds the C++ code from the data model).  The generated code ends up in the dom/bindings subdirectory of the objdir (e.g. dom/bindings/URLBinding.cpp).  There's some user-facing (as in, for people writing C++ code that's supposed to implement web stuff) documentation at https://developer.mozilla.org/en/Mozilla/WebIDL_bindings

Comment 10

[Mike Hoye [:mhoye]]

Can you confirm that you're still working on this bug?

Comment 11

[Boris Zbarsky [:bzbarsky]]

Attachment: Implement basic support for [TreatUndefinedAs=Missing] in WebIDL. does not add support for [TreatUndefinedAs=Missing] in overloaded

methods, since it's not clear what that should really look like.

This also does not require that [TreatUndefinedAs=Missing] on an
argument force it on all later arguments, since it's not clear that we
want that long-term.

Comment 12

[Boris Zbarsky [:bzbarsky]]

Stealing, because this is blocking people.

Comment 13

[Kyle Huey (Exited; not receiving bugmail, old account, do not use)]

Comment on attachment 755016 [details] [diff] [review]
Implement basic support for [TreatUndefinedAs=Missing] in WebIDL.   does not add support for [TreatUndefinedAs=Missing] in overloaded

Review of attachment 755016 [details] [diff] [review]:
-----------------------------------------------------------------

::: dom/bindings/parser/WebIDL.py
@@ +406,5 @@
>                          raise WebIDLError("[TreatUndefinedAs=Missing] is only "
>                                            "allowed on optional arguments",
>                                            [self.location])
> +                else:
> +                    if value == 'Null':

Is there a reason not to just do elif and not have a separate level of control flow.  I don't see one but it's possible I'm still too jetlagged to read code properly ...

Comment 14

[Boris Zbarsky [:bzbarsky]]

You're totally right.  I had the code looking a bit differently partway though writing the patch and then never reread it carefully enough...  Fixed to do elif.

https://hg.mozilla.org/integration/mozilla-inbound/rev/5f66c33dbf00

Comment 15

[Ed Morley [:emorley]]

https://hg.mozilla.org/mozilla-central/rev/5f66c33dbf00