387979 - Can't access about:blank iframe after reload

Status: RESOLVED FIXED

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

Description

[moz_bug_r_a4]

Attachment: testcase

This seems to be a regression from bug 381300.

http://bonsai.mozilla.org/cvsquery.cgi?module=AviarySuiteBranchTinderbox&branch=MOZILLA_1_8_BRANCH&date=explicit&mindate=2007-07-11+03&maxdate=2007-07-12+03

fx2.0.0.5pre 2007-07-11-03
Before reload: [object HTMLBodyElement]
After reload: [object HTMLBodyElement]

fx2.0.0.5pre 2007-07-12-03
Before reload: [object HTMLBodyElement]
After reload: Permission denied to get property HTMLDocument.body

Comment 1

[Boris Zbarsky [:bzbarsky]]

Looking into this...

Comment 2

[Boris Zbarsky [:bzbarsky]]

Attachment: Fix

This is basically a port to branch of the relevant parts of the fix for bug 337260

Comment 3

[Daniel Veditz [:dveditz]]

qawanted -- bug 337260 comment 6:
"Note to self: test the fix against bug 291745 and bug 292691" might be useful for regression testing (or not).

Comment 4

[Daniel Veditz [:dveditz]]

tried 5 different web mail systems (Gmail, yahoo classic and beta, Zimbra, and Open Webmail) without problems. Don't know who else is likely to use frames, reload them, and expect to reach into it.

Comment 5

[Martijn Wargers (dead)]

This can also affect designMode implementations, see:
http://www.mozilla.org/editor/midasdemo/
After a reload, you get a js error and you can't type anymore in the iframe.

I would strongly advise to get this patch on the branch...

Comment 6

[moz_bug_r_a4]

Also, http://www.squarefree.com/jsenv/ does not work after a reload.

Comment 7

[Daniel Veditz [:dveditz]]

Comment on attachment 272127 [details] [diff] [review]
Fix

r+ on the patch, clean port (plus branch-interface cruft).

I am waiting to give a+ until after I've actually tested the patch just in the rare, rare case bz's success with it depends on some other unnoticed local change.

Comment 8

[Daniel Veditz [:dveditz]]

Comment on attachment 272127 [details] [diff] [review]
Fix

carrying over sr= from the original patch.

Patch works on the regressions found so far: approved for the 1.8 branches, a=dveditz

Comment 9

[Johnny Stenback (:jst)]

Comment on attachment 272127 [details] [diff] [review]
Fix

sr=jst too fwiw

Comment 10

[Boris Zbarsky [:bzbarsky]]

Attachment: Patch actually checked in

This doesn't include the bits of bug 381300 in it.

Comment 11

[Daniel Veditz [:dveditz]]

I applied the original patch to a tree without 381300, then updated to get the last day of changes including 381300. The final tree that I ran my tests on matched exactly what bz actually checked in after the approvals.

Comment 12

[Boris Zbarsky [:bzbarsky]]

I checked in a mochitest for this.

Comment 13

[Boris Zbarsky [:bzbarsky]]

Oh, this is fixed on both branches as of this morning.

Comment 14

[Carsten Book [:Tomcat]]

verified fixed 1.8.1.5. with Mozilla/5.0 (Windows; U; Windows NT 5.2; en-US; rv:1.8.1.5) Gecko/20070713 Firefox/2.0.0.5 ID:2007071317

Works now as expected - also the midasdemo -> adding verified keyword