358670 - Make Firefox 2.0.0.x warn when I use nodes in the wrong document

Status: RESOLVED WONTFIX

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

Description

[Jesse Ruderman]

Trunk will soon throw WRONG_DOCUMENT_ERR when nodes are used in the wrong document, in order to comply with DOM specs and make interoperability with Safari easier.  (See bug 47903.)  Firefox 2.0.0.x should put a warning in the Error Console in the same situation, so extension and web developers will be more likely to notice their bugs before Firefox 3 comes out and breaks their code.

(Originally suggested by bz, bug 47903 comment 39.)

Comment 1

[Peter Van der Beken [:peterv]]

Yeah, I should have done this for FF 2.

Comment 2

[Peter Van der Beken [:peterv]]

Attachment: v1

Comment 3

[Jeff Walden [:Waldo]]

Comment on attachment 247570 [details] [diff] [review]
v1

>+    errorText.AppendASCII("Passing a node that has the wrong ownerDocument to "
>+                          "a DOM method. Future versions will throw "
>+                          "DOCUMENT_ERR exceptions, either create the node in "
>+                          "the right document or use adoptNode or importNode "
>+                          "to get a node with the right ownerDocument.");

This is a comma splice, I believe; how about:

Passing a node that has the wrong ownerDocument to a DOM method; future Gecko versions will throw DOCUMENT_ERR exceptions.  Either create the node in the right document or use adoptNode or importNode to get a node with the right ownerDocument.

(i.e., replace the period with a semicolon, uncapitalize "Future", change the comma after "exceptions" to a period, and capitalize "either"; I also added Gecko for greater clarity, but that's just my opinion that it's better that way)

Comment 4

[Mike Shaver (:shaver emeritus)]

Why is this WONTFIX?  Surely if the peril to developers of not knowing that their code is in violation of holy exception writ was worth risking major site breakage, it's still worth warning them about it.  s/"Future versions"/"Strictly specification-compliant implementations"/ and go?

Comment 5

[Peter Van der Beken [:peterv]]

Fine, I'm just pessimistic of the effect a warning like this will have.

Comment 6

[Dave Webber]

If this is still an option, I think it would be worth the experiment to see if we developers can be persuaded to move towards the spec (although perhaps rewording the warning might be warranted now). It's obviously rather too late for 2.x, but warnings won't break any sites, and another step away from browser-specific code could benefit web development and the ordinary user in the long run. Isn't part of the purpose of adopting web specifications to progress away from the demand on web developers to write several variations on numerous fragments (sometimes totalling a sizable portion) of web code?

Comment 7

[Jesse Ruderman]

Too late for Firefox 2.

Comment 8

[Dave Webber]

Understandable, but is there a definitive answer on whether this will ever be seen in 3.something, or has the prospect of trying to persuade web developers to move towards the spec on this been abandoned?

Comment 9

[Jesse Ruderman]

It was made to throw in bug 47903, but that was reverted in bug 418755.

Comment 10

[Dave Webber]

Yeah, I could see how throwing would cause excessive collateral damage, because of how long it was taken for granted that Firefox would not do so. What I was asking is, do you think it would be worthwhile at this point to simple add a warning or message in the console -- since web developers (presumably) watch it while testing their sites?

Comment 11

[Jesse Ruderman]

Given bug 418755 comment 28, seems unlikely.