289518 - Let nsXFormsUtils::ReportError() report fatal errors etc. too

Status: RESOLVED FIXED

(Core Graveyard :: XForms, defect)

Description

[Allan Beaufour]

As Aaron suggested in bug 287714, it should be possible to report fatal errors,
and not just messages, with nsXFormsUtils::ReportError(). biesi suggested we use
something like:
http://lxr.mozilla.org/seamonkey/source/js/src/xpconnect/idl/nsIScriptError.idl#101

Comment 1

[Doron Rosenberg (IBM)]

The main problem with nsIScriptError is that it always shows line numbers, even
if it is "0".

So the scripterror XBL binding needs some changing.

Comment 2

[Doron Rosenberg (IBM)]

The console issue is bug 289926 (xpfe) and 289927 (toolkit).

Comment 3

[Doron Rosenberg (IBM)]

The console issue is bug 289926 (xpfe) and 289927 (toolkit).

Comment 4

[Doron Rosenberg (IBM)]

Attachment: initial patch

Comment 5

[aaronr]

Comment on attachment 180382 [details] [diff] [review]
initial patch

+	 nsXFormsUtils::ReportError(NS_LITERAL_STRING("schemaLoadError"),
mElement, 0);

Need to use some kind of flag or define.  0 means nothing to me when reading
through the code.  Maybe use the same flag(s) that already exist in
nsIScriptError?

+	 ds->SerializeToString(tmpNode, srcLine);

I'd suggest using a different variable name.  This is more of a context type of
value.	When I see srcLine, I think the line number in the src or the actual
line of source.  Not serialized node info.


+   * @param aErrorType        Type of error.
+				 0 - error
+				 1 - warning
+				 2 - exception
+				 3 - strict

when you change to flags, make sure that you change the comments for
ReportError in nsXFormsUtils.h

Comment 6

[Doron Rosenberg (IBM)]

Attachment: with review comments addressed

Comment 7

[aaronr]

Comment on attachment 180400 [details] [diff] [review]
with review comments addressed

More that I think about it, why not just make the default error flag be
nsIScriptError::errorFlag?  Then most of these changes aren't necessary.  Could
even create a corresponding ::ReportWarning with a default of
nsIScriptError::warningFlag

But the code as it is has no problems that I can see.  Either way,  r=me

Comment 8

[Allan Beaufour]

(In reply to comment #7)
> (From update of attachment 180400 [details] [diff] [review] [edit])
> More that I think about it, why not just make the default error flag be
> nsIScriptError::errorFlag?  Then most of these changes aren't necessary.  Could
> even create a corresponding ::ReportWarning with a default of
> nsIScriptError::warningFlag

I second to that.

Comment 9

[Doron Rosenberg (IBM)]

Attachment: use defaults

Comment 10

[Allan Beaufour]

Comment on attachment 180486 [details] [diff] [review]
use defaults

Except for the bonus nop-change in nsXFormsUtils.cpp, everything's good.

Comment 11

[Allan Beaufour]

Checked in.