WSS4J interceptor creates creates invalid digest for SAAJ messages [SWS-345]

[gregturn]

Arjen Poutsma opened SWS-345 and commented

When using SAAJ and WSS4J on for outgoing messages, the digest created is incorrect, which results in signature verification failures. This is due to the fact that the Document exposed via SOAPMessage.getSOAPPart() is out of sync. Calling SOAPMessage.saveChanges() fixes this.

---

Affects: 1.5

Referenced from: commits 89f054f, 43eb72f

[gregturn]

Arjen Poutsma commented

As a workaround, you can use the AxiomSoapMessageFactory.

[gregturn]

Arjen Poutsma commented

Fixed.

[gregturn]

Patrick Haren commented

Arjen,

I am still seeing signature verification problems with version 1.5.1 in combination with SaajSoapMessageFactory.
The AxiomSoapMessageFactory workaround continues to work okay.
I am using spring-ws-1.5.1.jar from 04 May 2008, 1:47PM.

[gregturn]

Arjen Poutsma commented

What kind of problems, i.e. what is the exception and/or fault that you get?

[gregturn]

Arjen Poutsma commented

Definitely something weird going on here. The test I wrote for this (testSignAndValidate() in SaajWss4jMessageInterceptorSignTest) works fine, but it does not appear to work in deployed applications. Reopening for 1.5.2.

Let's hope we fix this one for good this time :(.

[gregturn]

Arjen Poutsma commented

Ok, I've given it another shot, with a different approach this time. I've also improved the tests, and tried this approach in a test project. It all works for me.

Patrick, could you try a snapshot tomorrow? See http://static.springframework.org/spring-ws/site/downloads/snapshots.html

[gregturn]

Patrick Haren commented

Arjen,
I just tried the snapshot (spring-ws-1.5.2-20080505.174403-1.zip) with a SaajSoapMessageFactory bean. Looks like it is working okay.
FYI, my test is against a WebSphere 6.1 web-service, using WebSphere's jax-rpc runtime, so the interoperability is looking good!

[gregturn]

Arjen Poutsma commented

Great! Thanks for checking this out, Patrick.

I guess it makes sense to speed up the release for this fix, so I've set 1.5.2 to be released at the end of next week.