124427 - targets are mixed if IFRAME in a FRAMESET loads source by JavaScript

Status: RESOLVED FIXED

(Core :: DOM: Navigation, defect)

Description

[mk]

From Bugzilla Helper:
User-Agent: Mozilla/5.0 (Windows; U; WinNT4.0; en-US; rv:0.9.8+) Gecko/20020208
BuildID:    2002020803

I made a very simple test case to show this bug:

- there's a frameset with two frames, upper and bottom.
- inside one frame is an iframe.
- the iframe's src is loaded by JavaScript onLoad of the frame.

First thing that's strange is that the browser doesn't stop loading.

Second, if you hit reload now, the iframe's content is loaded in the upper frame
(top.frames[0]) instead of the iframe (top.frames[1].frames[0]).

Reproducible: Always
Steps to Reproduce:
1. load test case http://www.rpgstudies.net/mozilla/
2. hit reload.


Actual Results:  iframe's content is reloaded into a frame.

Expected Results:  iframe's content should be reloaded into the iframe, frame
content into the frame.

If you go to another page and hit "back", the same happens. That makes me think
it's a problem with history.

Also after reloading once it takes ages to reload again.

Comment 1

[Boris Zbarsky [:bzbarsky]]

session history.  Seeing this on Linux too.

Comment 2

[mk]

Maybe it's got something to do with the missing src-attribute in the IFRAME 
element. The HTML 4.01 specification doesn't tell if it is required or not. On 
my page I have no initial value, so I didn't include it.

Everything seems to be alright when I add a src-attribute with a value.

When I use an src-attribute with a blank value, the browser doesn't stop 
loading, but doesn't mix the targets onReload either.

Comment 3

[mk]

> Everything seems to be alright when I add a src-attribute with a value.

That's only true if the src-value has the same value as the javascript-loaded 
src. Any other value results in the browser not stopping to load and the 
described frame defect when hitting the reload button.

Comment 4

[Radha on family leave (not reading bugmail)]

The patch attached to bug 105299 takes care of the iframe mismatch problem that
happens when you hit the reload button or go back forward to it. But When you
hit reload or go back, multiple history entries get created for the page in
session history. Shall investigate that. 

Comment 5

[Radha on family leave (not reading bugmail)]

Attachment: patch to docshell.

The patch looks specifically into cases where a subframe load could have been
initiated by  onLoadHandler and accordingly changes the loadType of teh
subframe, so that session hitory does not get confused. This is the case in
this testcase as well as the one in 88684.

Comment 6

[Adam Lock]

Comment on attachment 71013 [details] [diff] [review]
patch to docshell.

r=adamlock

Comment 7

[Alec Flett]

Comment on attachment 71013 [details] [diff] [review]
patch to docshell.

sr=alecf

Comment 8

[Asa Dotzler [:asa]]

Comment on attachment 71013 [details] [diff] [review]
patch to docshell.

a=asa (on behalf of drivers) for checkin to 0.9.9 and the mozilla trunk.

Comment 9

[Radha on family leave (not reading bugmail)]

Fix checked in.