85572 - Javascript Error for each message sent

Status: RESOLVED FIXED

(Other Applications :: ChatZilla, defect)

Description

[basic]

Build 2001061120 win32 installer sea talkback
This javascript error appears in the Javascript console for each message sent 
by chatzilla:

Error: [Exception... "'Stream would block' when calling method: 
[nsIStreamProvider::onDataWritable]"  nsresult: "0x80470007 
(NS_BASE_STREAM_WOULD_BLOCK)"  location: "<unknown>"  data: no]

Comment 1

[Robert Ginda]

The way the particular API that chatzilla uses to send messages requires that we
throw an exception to indicate we are done sending data.  In debug builds,
throwing an exception from a js component produces the debug message you're
seeing.  So, there's nothing going wrong, but the warning can get a bit annoying.

  The attached patch will keep the debug message from showing up for the
NS_BASE_STREAM_WOULD_BLOCK exception.

Looking for r/sr=

Index: src/xpcwrappedjsclass.cpp
===================================================================
RCS file: /cvsroot/mozilla/js/src/xpconnect/src/xpcwrappedjsclass.cpp,v
retrieving revision 1.50
diff -u -r1.50 xpcwrappedjsclass.cpp
--- xpcwrappedjsclass.cpp       2001/05/18 23:35:03     1.50
+++ xpcwrappedjsclass.cpp       2001/06/14 01:01:27
@@ -75,7 +75,8 @@
 // this function is factored out to manage that.
 static inline JSBool IsReportableErrorCode(nsresult code)
 {
-    return NS_FAILED(code) && code != NS_ERROR_FACTORY_REGISTER_AGAIN;
+    return NS_FAILED(code) && code != NS_ERROR_FACTORY_REGISTER_AGAIN &&
+        code != NS_BASE_STREAM_WOULD_BLOCK;
 }
 

Comment 2

[John Bandhauer]

Attachment: perhaps better patch

Comment 3

[John Bandhauer]

I agree with what rginda is doing. His is a nice minimal change. I'm proposing 
something a little different (but functionally equivalent). I'm thinking it 
is better to make it clearer how this list of nsresults might grow by using a 
switch statement (and also be more efficent if the list gets big). 

The checkin process downside is that you'll need to get someone else to sr= it. 
vidur?

Comment 4

[vidur (gone)]

sr=vidur. I agree with John, error codes that aren't really errors are a Bad Thing.

Comment 5

[Robert Ginda]

r=rginda on jband's patch.  Asa, will you a=?

Comment 6

[Asa Dotzler [:asa]]

a= [e-mail] for checkin to the trunk.
(on behalf of drivers)

Comment 7

[Robert Ginda]

checked in.