526709 - LS_QueryInterface throws the wrong exception (NS_ERROR_NOT_IMPLEMENTED) instead of (NS_ERROR_NO_INTERFACE)

Status: RESOLVED FIXED

(Toolkit :: Places, defect)

Description

[John J. Barton]

Every time I run firefox I get an uncaught exception error in 
file:///C:/mozilla/firefox-3.6b1pre.en-US.win32/firefox/modules/utils.js
line 128.
The problem is not on that line however. The bug is in nsLivemarkService.js.

I set statement
  debugger;
on line 842, the first line of LS_QueryInterface. Then run Firefox and Chromebug breaks on line 842. By single stepping I can see the service creation pass through LS_QueryInterface a number of times. Eventually it gets call with aIID
{986c11d0-f340-11d4-9075-0010a4e73d9a}
which is nsIClassInfo. On that call the QI issues 
    throw Cr.NS_ERROR_NOT_IMPLEMENTED; // 2147500033

Based on this I am pretty sure that the uncaught exception is caused by this throw. Most users do not see this because the XPCOM layer does not report exceptions for some (dubious) historical reason.

However to debug extensions we need to watch for all exceptions and this one is just noise, in addition to being a bug, but consequences I am uncertain.

So either livemark-service;2 needs to implment nsIClassInfo or the getService needs to not ask for it.

See also Bug 499568

Comment 1

[timeless]

Attachment: patch

Comment 2

[Marco Bonardo [:mak]]

Comment on attachment 410483 [details] [diff] [review]
patch

while this is correct, i'll file a bug to convert it to use XPCOMUtils unless someone has anything against that.

Comment 3

[John J. Barton]

I just want to check: I guess this change will cause getService to trap the throw? That is, this will fix the problem I see, not just change the error number?

Comment 4

[Marco Bonardo [:mak]]

GetInterface and QueryInterface should never report a NS_ERROR_NO_INTERFACE but will report others.

Comment 5

[John J. Barton]

Requesting blocking so that this issue is evaluated, given that the call occurs on every startup of firefox and the fix is small.

Comment 6

[Marco Bonardo [:mak]]

filed bug 526777 about XPCOMUtils

this probably doesn't block, but once it's on trunk will be easy to get approval.

Comment 7

[Dão Gottwald [:dao]]

http://hg.mozilla.org/mozilla-central/rev/7dac283e2028

Comment 8

[Marco Bonardo [:mak]]

notice in bug 526777 i've moved this code to XPCOMUtils, in case that would cause problems please contact me.

that means this code has been replaced on trunk, but we should take this patch on branches.

Comment 9

[Johnathan Nightingale [:johnath]]

Sucks, will gladly approve the fix, but wouldn't hold the release of Firefox 3.6 if this were the last bug.

Comment 10

[Dão Gottwald [:dao]]

http://hg.mozilla.org/releases/mozilla-1.9.2/rev/4ca2f1951447

Comment 11

[Marco Bonardo [:mak]]

Comment on attachment 410483 [details] [diff] [review]
patch

i suppose would be cool and not too much of a problem getting this on 1.9.1.7

Comment 12

[Daniel Veditz [:dveditz]]

Comment on attachment 410483 [details] [diff] [review]
patch

Approved for 1.9.1.8, a=dveditz for release-drivers

Comment 13

[Marco Bonardo [:mak]]

http://hg.mozilla.org/releases/mozilla-1.9.1/rev/459c9734217d