491553 - [meta] Keep Firebug and Chromebug working on 1.9.2

Status: RESOLVED FIXED

(Firefox :: General, defect)

Description

[John J. Barton]

This bug is to gather and track issues related to Firebug and Chromebug working on trunk

Comment 1

[John J. Barton]

I don't now how to express this in the bugzilla way, but 49152 prevents me from working on trunk.

Comment 2

[Rob Campbell [:rc] (:robcee)]

mind if we change this to 1.9.2 to be more specific? Otherwise, this could become a very long bug overtime.

Comment 3

[John J. Barton]

I got some help on Bug 491552 and Firebug 1.5 is fixed for the file list button.

Comment 4

[Rob Campbell [:rc] (:robcee)]

(In reply to comment #1)
> I don't now how to express this in the bugzilla way, but 491552 prevents me from
> working on trunk.

you should add those bugs to the "Depends on" field in this bug.

Comment 5

[Rob Campbell [:rc] (:robcee)]

johnj: why did you remove this bug if it's blocking? Even if it's been fixed, these typically stay in that field to show a historical record of progress. It is "the bugzilla way".

Comment 6

[John J. Barton]

(In reply to comment #5)
> johnj: why did you remove this bug if it's blocking? 

Please see comment 3 and bug 491552. Fixed or not bug 491552 does not block progress on Firebug 1.5 for FF 3.6.

Comment 7

[John J. Barton]

Firebug & Chromebug 1.5 come up on 
Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.2a1pre) Gecko/20090508 Minefield/3.6a1pre
But the FBTests don't run because the httpd.js server does not respond. I'll look into that after we merge 1.4 changes back into 1.5.

Comment 8

[John J. Barton]

Ok trying again to get Firebug up on FF3.6

We get "uncaught exception 2147500036", but on the C++ call stack it looks like the problem is around onInputStreamReady, nsIInputStreamCallback

When we come into nsresult
nsXPCWrappedJSClass::CheckForException, we have aPropertyName=onInputStreamReady
is_js_exception == 1
!xpc_exception is true
we enter XPCConvert::JSValToXPCException. But the second argument 's' resolves to a number. That 's' is bound to js_exception that came out of:
JSBool is_js_exception = JS_GetPendingException(cx, &js_exception);
That method is just a global-gettter; someone is setting the pending exception to an integer ?

If I track the throw events I see
onThrow from tag:11716:chrome://fbtest/content/httpd.js@1119: 236
and line 1119 is in the method onInputStreamReady.  Line 1119 is
         this._processBody();

So at this point there is yet another bug in the error handling and some problem in httpd.js reading the request

Comment 9

[John J. Barton]

After re-writing the error handling in httpd.js I discovered the 1.9.1 httpd.js will not work in 1.9.2.

After updating to the 1.9.2 httpd.js, FBTests run about half way with some errors. Then the process hangs on issue 1693, no CPU.  Reproduces.

Comment 10

[John J. Barton]

I built today:
Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.2a1pre) Gecko/20090728 Minefield/3.6a1pre

and I worked on changing FBTest to run off external servers (not quite done). But I did have the Firebug code up all day without obvious problem on FF3.6. Hopefully I can run the FBTests tomorrow.

Comment 11

[John J. Barton]

Updated FBTest to use an external server for test cases, tried with eclipse server and http://getfirebug.com/tests/content/.

Some FBTest pass on FF3.6; some fail. Some of the failures could be related to the test server URL changes I just made.

Comment 12

[Rob Campbell [:rc] (:robcee)]

see bug 501698 for progress on getting getfirebug.com/tests revved to support this. SVN update time's been reduced to 20 minutes and oremj's going to look into PHP upgrade.

Comment 13

[John J. Barton]

Ok we finally got  Firebug 1.5a21 tested under FF 3.5.2. So here is my first serious try with 3.6. DEBUG build, with patch from bug 508562

Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.2a2pre) Gecko/20090811 Minefield/3.6a2pre

On start we an assert: 
###!!! ASSERTION: sqlite3_close failed. There are probably outstanding statements that are listed above!: 'srv == SQLITE_OK', file g:/mozilla/mozilla-central/src/storage/src/mozStorageConnection.cpp, line 522

Firebug opens no errors. FBTest console opens and works.

Tests in the Firebug category pass. (activation/open/close/ new window)
Tests in Script category pass except 1575. It ends on a time out then we get multiple asserts:

###!!! ASSERTION: XPConnect is being called on a scope without a 'Components' property!: 'Error', file g:/mozilla/mozilla-central/src/js/src/xpconnect/src/xpcwrappednativescope.cpp, line 786

Then a crash, crash reporter does not come up.

When I rerun it, I get a lot more asserts. I think that the timeout is allowing a window.closed to hang around.So it may not be the test-code itself that causes the asserts.

Need to retry on nightly build.

Comment 14

[John J. Barton]

Firebug 1.5a21 on Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.3a1pre) Gecko/20090817 Minefield/3.7a1pre

One test fails, [FAILED] script/1575/issue1575Driver.js: New watch expression can't be created when breakpoint is active
- TIMEOUT: 10s [ERROR]

Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.2a1) Gecko/20090806 Namoroka/3.6a1

Same result.

Comment 15

[John J. Barton]

I realized in re-reading my last post I should say:

Firebug 1.5a21 passes almost all FBTest on FF 3.6 and 3.7. Only one case fails.