781154 - JS_ObjectIsDate returns false for Date objects created in a script.

Status: RESOLVED FIXED

(Core :: JavaScript Engine, defect)

Comment 1

[:Benjamin Peterson]

Test case?

Comment 2

[Mounir Lamouri (:mounir)]

Raphael, maybe you could attach a patch that adds the usage of JS_ObjectIsDate() for .valueAsDate and a test case? We need to have the other patches to land though.

Comment 3

[Raphael Catolino (:raphc)]

Attachment: input date base

Comment 4

[Raphael Catolino (:raphc)]

Attachment: input date sanitize

Comment 5

[Raphael Catolino (:raphc)]

Attachment: input date valueasdate

Comment 6

[Raphael Catolino (:raphc)]

Actually this problem only happens in the web console.
I saw this problem when testing the input <type=date> valueAsDate attribute. When you apply these patches (and enable the dom.experimental-forms setting) you should be able to use >input.valueAsDate = new Date().

The mochitests for valueAsDate() are working fine. But if you do the same in a console, it doesn't work, because JS_ObjectIsDate() returns false.

Comment 7

[:Benjamin Peterson]

It's probably failing on the wrapped version.

Comment 8

[Raphael Catolino (:raphc)]

I saw that the class of this object when created in the console was proxy instead of date. Is that what you call wrapped version? Can that happen somewhere else than in the console?
This behavior is disturbing for webdevs who want to use the web console to test code related to valueAsDate, but otherwise not a major issue for the input date implementation itself.

Comment 9

[Mounir Lamouri (:mounir)]

Very likely this bug can't be reproduced in Nightly any more because we use WebIDL bindings for input elements so maybe it could be interesting to have a look before it is no longer reproducible in all versions of Firefox?

Comment 10

[Jeff Walden [:Waldo]]

I'm not quite sure what's desired here.  Having the relevant JS_ObjectIsDate bits via WebIDL should solve this, yes, and when I test in a nightly it does indeed work.  (Although I ran across bug 874640 when doing so, and got derailed for a bit on that.)  What do you want me to do here, exactly?

Comment 11

[Mounir Lamouri (:mounir)]

(In reply to Jeff Walden [:Waldo] (remove +bmo to email) from comment #10)
> I'm not quite sure what's desired here.  Having the relevant JS_ObjectIsDate
> bits via WebIDL should solve this, yes, and when I test in a nightly it does
> indeed work.  (Although I ran across bug 874640 when doing so, and got
> derailed for a bit on that.)  What do you want me to do here, exactly?

I was wondering if you could look at this bug to see if it wasn't coming from a more important underlying bug that we might want to fix.

Comment 12

[Jeff Walden [:Waldo]]

Oh.  The underlying issue was all but certainly fixed in http://hg.mozilla.org/mozilla-central/rev/f8a7c4c4874d which would make us good now.

Comment 13

[Mounir Lamouri (:mounir)]

Thanks Waldo :)