262998 - Setting designMode doesn't work anymore in this particular case

Status: VERIFIED FIXED

(Core :: DOM: Editor, defect, P3)

Description

[Martijn Wargers (dead)]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.8a5) Gecko/20041003 Firefox/0.9.1+
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.8a5) Gecko/20041003 Firefox/0.9.1+

I'll attach a testcase.

The testcase works in:
Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.8a4) Gecko/20040911
Firefox/0.9.1+ (8:44 am)
But it doesn't work anymore in:
Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.8a4) Gecko/20040913
Firefox/0.9.1+ (9:46 am)


Reproducible: Always
Steps to Reproduce:
1. See testcase
2.
3.

Actual Results:  
No designmode for iframe. Instead I get an error in the js console:
Error: uncaught exception: [Exception... "Component returned failure code:
0x80004005 (NS_ERROR_FAILURE) [nsIDOMNSHTMLDocument.designMode]"  nsresult:
"0x80004005 (NS_ERROR_FAILURE)"  location: "JS frame ::
file:///C:/Documents%20and%20Settings/Martijn/Desktop/dm.htm :: dm :: line 20" 
data: no]

Expected Results:  
Set the designmode to on for the iframe without complaints.

Normally this doesn't work in any build when I just put some string in the iframe.
But before 2004-09-11, this did work because I used the
iframe.contentDocument.innerHTML=textarea.value (instead of
iframe.contentDocument.innerHTML='some random string').
Now that doesn't work also anymore.

I guess the fundamental problem is that I have display:none set to the iframe.
Setting designMode to an iframe with display:none is not possible (gives the
error message I mentioned previously).
But in the testcase, I set the display of the iframe back to block, just before
setting designMode to on. That doesn't work also normally: setting designMode
still would give the error message. But before 2004-09-11, the
iframe.contentDocument.innerHTML=textarea.value way seemed to work, but that
doesn't work nowadays.

Comment 1

[Martijn Wargers (dead)]

Attachment: Testcase

Sorry for the rather complicated bug report. 
I think it would just better if it would be possible to set designMode on
display:none iframes.
This bug is just a side issue of that, I guess (some strange workaround which
worked, but nowadays not anymore.

Comment 2

[Boris Zbarsky [:bzbarsky]]

> Setting designMode to an iframe with display:none is not possible

That's pretty bogus.

jst, is there a reason this code tries to go through prescontext to get the
window instead of just QIing nsIScriptGlobalObject?

Comment 3

[Boris Zbarsky [:bzbarsky]]

Attachment: Somewhat like this, perhaps?

Comment 4

[Johnny Stenback (:jst)]

(In reply to comment #2)
> > Setting designMode to an iframe with display:none is not possible
> 
> That's pretty bogus.
> 
> jst, is there a reason this code tries to go through prescontext to get the
> window instead of just QIing nsIScriptGlobalObject?

Nope, at least no other reason than historic ones, I guess...

Comment 5

[Johnny Stenback (:jst)]

Comment on attachment 161284 [details] [diff] [review]
Somewhat like this, perhaps?

r+sr=jst

Comment 6

[Boris Zbarsky [:bzbarsky]]

Fixed.

Comment 7

[Martijn Wargers (dead)]

So this should be fixed in the build I'm using, right?
Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8a5) Gecko/20041007
Firefox/0.9.1+ (8:21 am)
The testcase still doesn't work for me.

The URL testcase is the thing that I hope that gets fixed by this bug.
Setting the designmode on second, third, etc instance doesn't work in that case.

Comment 8

[Martijn Wargers (dead)]

Attachment: Testcase2

This testcase is showing more explicitly that setting designMode is not working
for iframes with display:none.

Comment 9

[Boris Zbarsky [:bzbarsky]]

Ok, reopening.  I can fix the original testcase here (by flushing pending
restyles), but the bogosity of not being able to set designMode on display:none
iframes will remain -- nsEditingSession::SetupEditorOnWindow requires a presshell.

Comment 10

[Boris Zbarsky [:bzbarsky]]

Attachment: Ensure frames are up to date before grabbing for presshells

This fixes "Testcase" by ensuring that the display change is processed before
we look for a presshell, but not "Testcase2" (since there is simply no
presshell in site there).  Martijn, would you file a separate bug on that,
please?

Also, with this patch you don't need the textarea.value hack at all.

Comment 11

[Johnny Stenback (:jst)]

Comment on attachment 161389 [details] [diff] [review]
Ensure frames are up to date before grabbing for presshells

r+sr=jst

Comment 12

[Martijn Wargers (dead)]

(In reply to comment #10)
> Martijn, would you file a separate bug on that,
> please?
Ok, I filed bug 263392 on that.

Comment 13

[Boris Zbarsky [:bzbarsky]]

Fixed again, for real.  ;)

Comment 14

[Martijn Wargers (dead)]

(In reply to comment #13)
> Fixed again, for real.  ;)
Indeed. Thanks Boris!

Comment 15

[Boris Zbarsky [:bzbarsky]]

Pushed testcase: http://hg.mozilla.org/mozilla-central/rev/39ce9c1e4b56