On 08.08.2005 06:16:58 Manuel Mall wrote:
> On Mon, 8 Aug 2005 01:04 am, Jeremias Maerki wrote:
> <snip/>
> >
> > Another important area would be URI resolution and proper and
> > consistent resolution of relative paths. I think that is something
> > that bugs especially our users through Cocoon. There are a few
> > (older) notes about that in the Wiki. In this context it might be
> > worthwhile to have a closer look at Batik's ParsedURL which I'd wish
> > to see in XML Graphics Commons and used in FOP if that makes sense.
> > This area could be something very very useful for the project and its
> > users and be isolated enough for you not to get lost in FOP's
> > complexity. Still, it's not a minor bit to chew. Anyway, it's up to
> > you.
> >
> Jeremias,
> Happy to look into this. Had a quick look at ParsedURL in Batik. Looks 
> reasonably straightforward as it is a replacement for java.net.URL. 
> That would solve the idea of being able to support custom protocols. 

We are able to do that already but with the known constraints of what
the JDK provides.

> But the other issue you mention is "proper and consistent resolution of 
> relative paths". Can you elaborate for my benefit under which 
> circumstances relative path resolution in fop is broken as I don't 
> quite understand where the problem is.

Yes, sorry. I don't have all the details present as I investigated this a
few months ago (and didn't write the results down, d'oh). Basically,
I've stumbled upon a few anomalies with relative paths and resource
access in general. Images were not properly loaded and I think there
were differences in behaviour between FOP and Batik. Something like that.
What I'd like would be a set of test cases (JUnit-based) that tests all
sorts of resource references (absolute and relative URIs, normal
filenames, for images (bitmap formats and SVG tested separately),
references to images inside SVG code to see if the Batik behaves
identically in our setup etc. It should also be verified if
FOUserAgent.getStream() works as expected and, when done, if the URI
resolution mechanism works as expected. Stuff like that. Not all very
exciting but quite critical for user experience. It might well be that
you end up improving the image subpackage along the way. It's less than
perfect ATM. See this example:

So, a perfect playground. :-)

Jeremias Maerki

Reply via email to