865289 - Can't open exception source from the console

Status: RESOLVED INCOMPLETE

(DevTools :: Debugger, defect, P3)

Description

[Mihai Sucan [:msucan]]

Boris points us to:

http://www.jango.com/stations/263448187/tunein?gcid=1&l=0

which displays a bunch of warnings and one TypeError:

[18:33:28.080] TypeError: _jui.jango_player is undefined @ http://cd09.s3.static.jango.com/javascripts/proto_player_sm2_37808.js:855

If you try to click the exception source link bz reports that the debugger opens and it fails to show the script source, with a nightly from the 19th of April. I cannot reproduce that error with my local fx-team build of today. I get the debugger opened for a very short time then view source opens. With today's nightly bz gets a crash. He also reports that yesterday's nightly shows the same behavior as that on my system.

After I open the debugger on that page (via click on message source) I get a dbg-script-actors.js exception:

[18:33:24.846] [Exception... "Component returned failure code: 0x80040111 (NS_ERROR_NOT_AVAILABLE) [nsIChannel.asyncOpen]"  nsresult: "0x80040111 (NS_ERROR_NOT_AVAILABLE)"  location: "JS frame :: chrome://global/content/devtools/dbg-script-actors.js :: fetch :: line 2695"  data: no]
undefined @ chrome://global/content/devtools/dbg-server.js:608

Followed by the client-side error:

[18:33:24.848] Server did not specify an actor, dropping packet: {"error":"unknownError","message":"error occurred while processing 'source': [Exception... \"Component returned failure code: 0x80040111 (NS_ERROR_NOT_AVAILABLE) [nsIChannel.asyncOpen]\"  nsresult: \"0x80040111 (NS_ERROR_NOT_AVAILABLE)\"  location: \"JS frame :: chrome://global/content/devtools/dbg-script-actors.js :: fetch :: line 2695\"  data: no]\nundefined"} @ resource://gre/modules/devtools/dbg-client.jsm:592

Conclusions until now:

1. fetch() is used to fetch sources and it fails to do so. Looks like a regression caused by source maps landing (this is new code to me, maybe there's a different root cause here).

2. the web console doesn't always handle gracefully cases when the debugger cannot display the script source.

Comment 1

[Boris Zbarsky [:bzbarsky]]

As far as I can tell, the behavior changed in http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=50d25e083421&tochange=acf388eaf9e9 to go from "show nothing" to "switch to debugger panel showing nothing, open view source, switch back to console".

Comment 2

[Boris Zbarsky [:bzbarsky]]

So I just tried building a debug build from rev acf388eaf9e9 and it has this problem, so the fact that it started working in a nightly is ... odd.

Comment 3

[Boris Zbarsky [:bzbarsky]]

Oh, and I get:

[14:07:41.104] TypeError: can't access dead object
TA__addDebuggees@chrome://global/content/devtools/dbg-script-actors.js:139
TA_findGlobals@chrome://global/content/devtools/dbg-script-actors.js:153
TA_onAttach@chrome://global/content/devtools/dbg-script-actors.js:222
DSC_onPacket@chrome://global/content/devtools/dbg-server.js:655
LDT_send/<.run@chrome://global/content/devtools/dbg-transport.js:224
 @ chrome://global/content/devtools/dbg-server.js:608
[14:07:41.132] Server did not specify an actor, dropping packet: {"error":"unknownError","message":"error occurred while processing 'attach': TypeError: can't access dead object\nTA__addDebuggees@chrome://global/content/devtools/dbg-script-actors.js:139\nTA_findGlobals@chrome://global/content/devtools/dbg-script-actors.js:153\nTA_onAttach@chrome://global/content/devtools/dbg-script-actors.js:222\nDSC_onPacket@chrome://global/content/devtools/dbg-server.js:655\nLDT_send/<.run@chrome://global/content/devtools/dbg-transport.js:224\n"} @ resource://gre/modules/devtools/dbg-client.jsm:592

Comment 4

[Boris Zbarsky [:bzbarsky]]

That is, I get that in a debug build from rev acf388eaf9e9.

Comment 5

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

Latent bug.

Comment 6

[Eddy Bruel [:ejpbruel]]

Just guessing here, but could this be related to bug 842937 in some way?

Comment 7

[Jason Laster [:jlast]]

is this a duplicate of some of the other console links bugs?

Comment 8

[Brian Hackett [Laid off!]]

(In reply to Jason Laster [:jlast] from comment #7)

is this a duplicate of some of the other console links bugs?

It's hard to say, the STR doesn't work anymore, not surprisingly. Bug 1447244 should be able to fix problems like this, but the main places I've seen problems today with invalid view-source links is with generated sources (eval etc.), which don't seem to be in use here.

Comment 9

[Jason Laster [:jlast]]

Honza, what do you think we should do here?

Comment 10

[Jan Honza Odvarko [:Honza] (always need-info? me)]

I can't reproduce the bug either (STR doesn't apply anymore)
Comment #3 could indicate and old e10s bug ('TypeError: can't access dead object'). I've seen a lot of those back then.

I am closing since there isn't a way to reproduce it.

Honza