630454 - Canvas defense game doesn't work in Minefield, throws "world is not defined"

Status: RESOLVED FIXED

(Tech Evangelism Graveyard :: English US, defect)

Description

[Matheus Kerschbaum]

User-Agent:       Mozilla/5.0 (Windows NT 6.1; rv:2.0b11pre) Gecko/20110131 Firefox/4.0b11pre
Build Identifier: Built from http://hg.mozilla.org/mozilla-central/rev/732a38102733

The canvas defense game (http://canvasdefense.appspot.com/) doesn't seem to work in Minefield, but works just fine in IE9 and Chrome.

The game is updated fairly often, and I remember the first time I played it did work in Minefield, so I don't know if this a regression or simply a bug that always existed but the game simply wasn't hitting it.

The game also fails to work regardless of TM/JM being turned on/off.

Reproducible: Always

Comment 1

[David Mandelin [:dmandelin]]

Confirmed in latest nightly.

Comment 2

[Jan de Mooij [:jandem]]

Attachment: Test case

This alerts "object" in Chrome, "undefined" in FF 4. It works if I remove the doctype, which makes it unlikely to be a JS engine bug.

Comment 3

[Mike Shaver (:shaver emeritus)]

Sounds like evang, if I understand the HTML5 interaction here correctly.  Henri?

Comment 4

[Boris Zbarsky [:bzbarsky]]

Actually, HTML5 specs this insane behavior for standards mode too, at the moment.

That said, I think that this is wrong and should change.

I assume the game is broken in Fx3.6 as well, right?

Comment 5

[Boris Zbarsky [:bzbarsky]]

And in particular, my assumption is right this shouldn't be a blocker.

Comment 6

[Henri Sivonen (:hsivonen)]

This isn't a parser bug. This is a DOM bug if we take the spec as being correct. This is an evang bug if we take the spec as being wrong.

IIRC, Brendan has previously argued pretty strongly that the spec is wrong here and Gecko should stay as-is.

Comment 7

[Boris Zbarsky [:bzbarsky]]

I filed http://www.w3.org/Bugs/Public/show_bug.cgi?id=11960

Comment 8

[Brendan Eich [:brendan]]

Brendan and all of Ecma TC39 and a bunch of Mozillans including jst all object to this.

The originally-IE-only pollution of the global scope by any id'ed DOM element is a botch, and we emulated it only in quirks mode to grow Firefox market share starting in 2004. Standardizing it for standards mode is a worse botch, and unnecessary in our experience.

But now we're playing Prisoner's Dilemma round 2 and up-and-coming browsers are emulating it in standards mode as well as quirks mode documents, misleading content authors to expect us to do so as well. Time for a better game.

/be

Comment 9

[Brendan Eich [:brendan]]

(In reply to comment #8)
> Brendan and all of Ecma TC39 and a bunch of Mozillans including jst all object
> to this.

jst and bz (I cited jst because he did the implementation work in 2004).

> But now we're playing Prisoner's Dilemma round 2 and up-and-coming browsers are
> emulating it in standards mode as well as quirks mode documents, misleading
> content authors to expect us to do so as well. Time for a better game.

So this should be a TE bug.

/be

Comment 10

[Boris Zbarsky [:bzbarsky]]

Brendan, did you mean to comment on the W3 bug?  ;)

Comment 11

[Johnny Stenback (:jst)]

Over to tech evang.

Comment 12

[Asa Dotzler [:asa]]

This should come off the hardblocker list since it is no longer classified as a Firefox bug. If you have the authority and you disagree, just add the whiteboard back but I'm taking it off for now.

Comment 13

[Christopher Blizzard (:blizzard)]

I've been talking to some google people.  This is an agency demo and they are trying to get it updated.

Comment 14

[Boris Zbarsky [:bzbarsky]]

Yep, they've fixed it.