Re: Hysteria

[Christian Geuer-Pollmann <ge...@nue.et-inf.uni-siegen.de>]

Hi Axl,

first of all, congrats and a big thanks for your effort. I didn't hear 
about this StreamSource prob anywhere. If you commit the code to CVS, 
there's a chance to have a look at the problem ;-)) Best way would be a 
unit test which runs on Linux and fails on windows. If you can isolate the 
test from the encryption code, you can also open a bugzilla bug for 
Xerces/Xalan/dunnoWhat.

Kind regards,
Christian

BTW, yesterday, I presented Apache xml-security at SUN tech days in 
Düsseldorf [1] and paised SUN (especially SUN Ireland and you) for backing 
Apache ;-))

[1] End-to-End Web Services Security
    <http://www.sun.com/developers/techdays/agenda/duesseldorf.html>






--On Dienstag, 22. Oktober 2002 16:41 +0100 Axl Mattheus 
<ax...@sun.com> wrote:

> I have completed an XML Encryption implementation for Apache. BUT, and
> this is a big but - it works fine on Linux, but not on Windows (I only
> discovered this last night as I started testing on different platforms as
> preparation to check it in to the Apache CVS repository).  I use the
> exact same version of the JDK (j2sdk1.4.0) - so, I assume that the write
> once run everywhere stuff applies...  I use the same libraries (bouncy
> castle (bcprov-jdk14-115.jar) for crypto, junit 3.8.1 for testing and ant
> 1.5 for building the source) - to be precise it is the same jar, I
> downloaded it and ftp'd it to my Windows box.  I use the same source
> (internal CVS repository).  The problem seems to lie with the
> javax.xml.transform implementation.  On Solaris and Linux it works fine
> if I transform a DOM node to a string (via a StreamSource) but not on
> Windows.  It seems that during the transform the output get corrupted.
> Has anybody encountered this anomoly?  Any help will be greatly
> appreciated.