[xsd-users] Roundtrip errors with CXX-Tree

[Russ Johns]

Thanks for the prompt reply-

it seems that it did not find the schema, though on Windows  I made sure
it was there-  when I switched to linux I added the schema::properties
coding,  but apparently forgot to copy the schema to the new development
directory.  Maybe there was a typo in the spelling of the schema
filename on windows...  I'll check.

It works fine now... at least in Linux on xsd 3.1...


On Tue, 2008-03-11 at 08:43 +0200, Boris Kolpackov wrote:
> Hi Russell,
> 
> Russell C. Johns <johns at lanl.gov> writes:
> 
> > terminate called after throwing an instance of
> > 'xsd::cxx::tree::parsing<char>'
> >   what():  instance document parsing failed
> 
> Try to catch this exception and print it as you did with the
> serialization case:
> 
> try
> {
>   std::auto_ptr<TEST::input::TestInput> tif
>     (TEST::input::TEST_input(argv[1],0,props));
> }
> catch (const xml_schema::exception& e)
> {
>   std::cerr << e << std::endl;
> }
> 
> This way you will get more information about the error.
> 
> 
> > I have tried this on windows and linux, with versions 2.3, 3.0, and 3.1
> > all produce the same error...
> >
> > is there an error in the parsing design?
> 
> There might be, or it could be the way you construct the object model
> from the text format (note also that XMLSpy is famous for successfully
> parsing invalid XML documents and schemas). If printing the error does
> not shed any new light, we will need a compilable test case to debug.

FYI-  I was thinking my parsing design rather than yours.   but I
figured that XSD's/ Xerces serialization validation would be as strong
as it's parsing validation.

There are one or two holes in XSD/ Xerces validation, though they are
not particularly problematic.  On one test element  that was restricted
by  ennumeration,  it did not catch that I had added a new value to the
ennumeration without updating the Schema.  Other than that I have not
observed any errors...

-Russ
> 
> Boris