781452 - when window.open is called mozbrowserloadstart and mozbrowserloadend are fired two times

Status: RESOLVED INVALID

(Firefox OS Graveyard :: General, defect)

Description

[Jose Manuel Cantera]

Attachment: popup_manager.js

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_6_8) AppleWebKit/537.3 (KHTML, like Gecko) Chrome/22.0.1221.0 Safari/537.3

Steps to reproduce:

open the iframe listen to mozbrowserloadstart, mozbrowserloadend


Actual results:

the mozbrwoserloadstart and loadend are called two times. someone said this is due to the about:blank initial loading

Comment 1

[Justin Lebar (not reading bugmail)]

> someone said this is due to the about:blank initial loading

Likely.

I'm not going to fix this anytime soon, because this should not block any work.

If you want to try to fix this in BrowserElementChild.js, it shouldn't be particularly hard to hack together, and I'm happy to review the patch.  Just don't fire the events when we're at about:blank and we've never been to another page.  You'll also have to fix the automated tests which work around these additional events.

Comment 2

[Gene Lian (I already quit Mozilla)]

We got Gaia developers complaining about this when designing the first-run wizard. Alive may elaborate more on this.

Comment 3

[(inactive after 6/18) Alive Kuo [:alive]]

FTU now calls window.close() to terminate itself on it finishes.
Gaia/System App/Window Manager will get two mozbrowserclose at the same time, which may cause some buggy behavior. I don't know if this is the same thing as this bug but I am going to workaround in gaia in bug 813541

Comment 4

[Hossain Al Ikram [:ikram] (QA Contact)]

Thanks for filing this bug. As the bug was filed a long days ago, I suppose it may have been resolved. Would you kindly check again and let me know , if you can re produce the bug Now or not. This thing will be really appreciated.

If you could re produce the bug , please let me the recent status of the bug.

Regards

Comment 5

[Jose Manuel Cantera]

I think :alive can respond better than me

Comment 6

[(inactive after 6/18) Alive Kuo [:alive]]

I am not seeing the issue for a long time. Maybe it's already fixed. Kanru are you able to confirm?

Comment 7

[Kan-Ru Chen [:kanru] (UTC+9)]

(In reply to Alive Kuo [:alive][NEEDINFO!] from comment #6)
> I am not seeing the issue for a long time. Maybe it's already fixed. Kanru
> are you able to confirm?

We have the protection code even before this bug was filed. I believe this case is covered by following tests:

  dom/browser-element/mochitest/test_browserElement_inproc_LoadEvents.html
  dom/browser-element/mochitest/test_browserElement_oop_LoadEvents.html

In short, these tests make sure we only fire loadstart/locationchange/loadend once and the locationchange is the targeted location instead of about:blank. So, WORKSFORME?

Comment 8

[Kan-Ru Chen [:kanru] (UTC+9)]

So the following sequence will produce two loadstart/loadend

  document.body.append(iframe); // first, load about:blank
  iframe.src = url;             // second, load url

And this produces only one loadstart/loadend

  iframe.src = url;
  document.body.append(iframe);

Comment 9

[Jose Manuel Cantera]

as per comment #8