Hi Berin,
No, because we really need to have a good design discussion. Maybe we
should start throwing some ideas around on the list and getting an idea of
people's requirements? One thing that's always bugged me about the
ProblemListener architecture is that it's very intrusive.
I think an architecture based on exceptions would be much simpler. Then,
we can always pass information to the application through a ProblemListener
when the exception is caught. That way, exceptions are thrown at the point
where the error occurs, and not through the ProblemListener. We can still
keep the ProblemListener for warnings, and informational messages.
So, everyone out there who is interested in error handling, let's hear from
you!
Dave
<[EMAIL PROTECTED]
om.au> To:
[email protected]
cc: (bcc: David N
Bertoni/Cambridge/IBM)
01/22/2003 05:56 Subject: Re: RE: Getting the
line and column number of an error
PM
David,
Out of curiosity, is there actually a plan to do
the overhaul? It seems to be something of
interest to a number of people?
Cheers,
Berin
>
> From: David N Bertoni/Cambridge/IBM <[EMAIL PROTECTED]>
> Subject: RE: Getting the line and column number of an error
> Date: 23/01/2003 3:25:02
> To: [email protected]
>
>
>
>
>
> Hi Don,
>
> There's no nightly build, but if you get the latest code from CVS and
build
> it yourself, you should see these coming through. Unfortunately, due to
> our convoluted error handling mechanism, they are coming through as
> warnings, and not errors. )Yet another reason why it needs to be
> overhauled.)
>
> Dave
>
>
>
>
> "Don McClimans"
> <[EMAIL PROTECTED] To:
<[email protected]>
> ronics.com> cc: (bcc: David
N Bertoni/Cambridge/IBM)
> Subject: RE: Getting
the line and column number of an error
> 01/21/2003 12:45 PM
>
>
>
>
> Dave,
>
> I'm back with more questions about error handling...
>
> I accidentally generated a source file with a 0x04 character in the text
of
> an element (UTF-8 encoding). Xalan (correctly) does not like this. When I
> run the command line xalan.exe, I get the message:
> Fatal Error at (file test.txt, line 15729, column 24):
Invalid
> character
> Unicode: 0x4)
> Invalid character (Unicode: 0x4)
>
> In my application, my custom problemListener is never called, and
> XalanTransformer::getLastError() returns:
> Invalid character (Unicode: 0x4)
>
> So I'm back to wondering how to get the line number of the error.
>
> I tried creating a std::ostringstream, and calling
> XalanTransformer::setWarningStream with it. Unfortunately, no message is
> stored in the ostringstream. Browsing the xalan code, this has me
confused,
> as it looks like that's where the first message ("Fatal error...") above
is
> coming from in the xalan.exe example. Should this work?
>
> I'm now running the 2002/10/21 interim release. Should I upgrade to the
> latest nightly build to (presumably) get your fix to return the line
number
> in getLastError()? How stable is this? I need something that I can
release
> to others.
>
> Or is there another way to get at that information?
>
> Thanks for any help you can give me.
>
> Don
>
>
>
>
This message was sent through MyMail http://www.mymail.com.au
