Philipp Chudinov wrote:
> :) Yes, we are in circles. No, I do not generate any XML-RPC elements
> from jsp or anything else. Well, here is waht i've found:
> 1) Xindice is a native xml bd
> 2) It is so killing stupid sometimes
> 3) That is why if you want to be happy, you will need XML-RPC
> 4) I hate the day i started to use it ;)
[snip]

OK, I've taken a look at Xindice and the XML-RPC interface. I think I
understand what's happening and it's nothing to do with JSP or UTF-8
encoding.

The Xindice XML-RPC interface retrieves the XML document as a Java String
and passes it back to the caller using the standard Apache XML-RPC
mechanism. The problem arises if the returned String has Unicode characters
with values > 255. Our current released implementation of XMLWriter uses a
Writer with an encoding of "8859_1". However we do not check that the
characters we are writing have Unicode values < 256. The Writer represents
such characters as '?'. This is why you see the result you do. The version
in CVS would barf on a value > 255.

What should happen is that XMLWriter sends characters > 255 as &xxxx; . I
will hack a version to do this and send it to you offline. The code will be
from the CVS version so you should probably check out the source from CVS as
I can't guarantee that it will work with the released version you have.

John Wilson
The Wilson Partnership
http://www.wilson.co.uk

Reply via email to