1036866 - JSF postback causes wrong debugger window content

Status: RESOLVED FIXED

(DevTools :: Debugger, defect)

Description

[Jakub Czerny]

JSF (Java Server Faces, Java EE web framework) postback causes that the same url may return different html code. I such cases Firefox shows wrong content in Debugger window. More specifically it keeps showing script that was served by certain url for the first time. Even though wrong script is shown firefox performs right code, so the problem "only" prevents meaningful from debugging.

The same problem is also observable in FireBug.

The was tested and appears in Firefox 30 and 33 (Nightly).

There is demo application running at http://jsfpostback-czerny.rhcloud.com/, source code at https://github.com/jakub-n/jsfpostback.

Steps to reproduces (fails for any site using postback without redirection):
1. Go to http://jsfpostback-czerny.rhcloud.com/
2. Check that source code (Ctrl-U), Debugger window, rendered page and console output matches.
3. Click the link "to page B".
4. Check that source code, rendered page and console output matches. However Debugger window content remain unchanged (FAIL).

Expected result:
Debugger window shows script that matches loaded page.

Comment 2

[Panos Astithas (he/him) [:past] (please ni?)]

I can reproduce this. James, does your Debugger.Source work fix this by any chance?

Comment 3

[James Long (:jlongster)]

I'm not sure, I haven't looked into how we show the loaded html page in the source view, and map that back to JS sources. But most likely my work will affect this for the better. I'll keep track of it and take a look at this when I get further along. If it doesn't happen to fix it, I can debug it more.

Comment 4

[Eddy Bruel [:ejpbruel]]

I bet you this is another instance of bug 893696.

Comment 5

[James Long (:jlongster)]

I'm far enough along in my Debugger.Source work to say that it won't fix this. We don't have the full HTML page as a source anywhere, and when it loads in the debugger have to fetch the whole thing again.

I'm sure we could figure out how to fix that (sending an extra request is annoying), but my patch will do the same thing.

Comment 6

[Eddy Bruel [:ejpbruel]]

(In reply to James Long (:jlongster) from comment #5)
> I'm far enough along in my Debugger.Source work to say that it won't fix
> this. We don't have the full HTML page as a source anywhere, and when it
> loads in the debugger have to fetch the whole thing again.
> 
> I'm sure we could figure out how to fix that (sending an extra request is
> annoying), but my patch will do the same thing.

Thanks for letting me know. Dropping the dependency on bug 905700.

My hunch is still that the underlying cause is the same as bug 893696, and fixing that would probably fix this issue as well.

Comment 7

[Jason Laster [:jlast]]

Closing as the demo no longer works. Jakub, if you still see the issue feel free to re-open.