[RFC][icedtea-web] Fix handling absolute paths passed into jnlp_href's value.

[Omair Majid]

On 06/04/2012 11:42 AM, Danesh Dadachanji wrote:
> Too late =) I wasn't too happy with this change either so I hunted down
> a better solution. I noticed some of the tests used JNLPs directly so I
> attempted to integrate that into my test and use the first patch
> (without the evaluation wrapper). However, Omair pointed out that using
> JNLPs is not ideal because this adds test coverage of the parser. So to
> avoid this, I've added the strategy class JNLPCreator that wraps around
> the creation of a JNLPFile. Under netx, it will simply do what it did
> before, create a normal JNLPFile. In the test however, it creates a mock
> JNLP, one that only has the bits my test in particular cares about (e.g.
> location). The additional class in the unit test is there to ensure
> PluginBridge's constructor can run without throwing exceptions.

Looks good to me.

Cheers,
Omair