253127 - xpconnect error report should be quieted for some results

Status: RESOLVED WONTFIX

(Core :: XPConnect, enhancement)

Description

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

In a few cases, a js object throwing an error should not cause "* Call to
xpconnect wrapped JSObject ..." spam.  nsIInterfaceRequestor's getInterface
method uses NS_NOINTERFACE (NS_ERROR_NO_INTREFACE?) to indicate that it can't
provide the requested interface.. this causes a lot of spam for, say,
nsIChannel's notificationCallbacks implemented in JS, as they're asked whether
they implement a number of different intrefaces.

Comment 1

[timeless]

vlad: i converted QI impls to use Components.returnCode. some people seemed to be willing to 
whitelist QI, but GI and others of course have the same behavior...

do you have a plan?

Comment 2

[Reed Loden [:reed]]

*** Bug 347310 has been marked as a duplicate of this bug. ***

Comment 3

[John J. Barton]

This bug gives a steady stream of bogus error messages that cannot be prevented. In my book that's more than "min" so I've changed it to encourage someone to fix it.

Comment 4

[Justin Dolske [:Dolske]]

Is this still an issue on trunk? I watch console output on my debug builds, and I don't see excessive amounts of output from thrown exceptions (although they do happen now and then).

Bug 393627 et al added some code to suppress thrown exceptions in some cases; I'm not clear if there was more work pending from that effort or not. Feels like we should either mark this as FIXED or break it out into the specific exception cases that need additional special-casing.

Comment 5

[John J. Barton]

Using a quick test based on http://code.google.com/p/fbug/issues/detail?id=434 with FF3.0b4pre, the messages are trapped by catch and don't come out on the consoleService.

Comment 6

[Myk Melez [:myk] [@mykmelez]]

Note that bug 393627 regressed reporting of errors in certain cases (see f.e. bug 415498), so I'd hesitate to resolve this bug fixed until those regressions get resolved and their fix doesn't reintroduce the issues here.

Comment 7

[Mike Shaver (:shaver emeritus)]

That issue has a separate bug, we can REOPEN this if it regresses.  Put a unit or mochi test in, and you'll be sure. :)  I'm boldly agreeing with Justin.

Comment 8

[Sergey Yanovich]

(In reply to comment #6)
> Note that bug 393627 regressed reporting of errors in certain cases (see f.e.
> bug 415498), so I'd hesitate to resolve this bug fixed until those regressions
> get resolved and their fix doesn't reintroduce the issues here.

I've filed a fix for bug 415498, that won't regress bug 393627. This may safely be dupped to bug 393627.

Comment 9

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

this is still showing "new". I'm marking RESO FIXED as per comments #4 and #7.

Comment 10

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

correction... ->WONTFIX