Aegis and unlabelled HashMap

[Benson Margulies <bi...@gmail.com>]

Aegis design question.

If you put a raw Set into an SEI, Aegis throws a complaint that it can't
figure out the type.

If you put a raw HashMap into an SEI (at least, via a complex JAX-WS
annotation I sent along yesterday), it cheerfully builds up some sort of
Any2AnyHashMap mapping, which discards a wide variety of input with no
comment whatsoever.

Is this a good thing? Should the raw HashMap result in a diagnostic like the
Set?

Re: Aegis and unlabelled HashMap

[Daniel Kulp <dk...@apache.org>]

On Sunday 24 February 2008, Benson Margulies wrote:
> Aegis design question.
>
> If you put a raw Set into an SEI, Aegis throws a complaint that it
> can't figure out the type.

I commented about this in your JIRA, but IMO, the Set case should be the 
same as the Map and List case where it would build up a AnySet type that 
is a set of xsd:anyType things.   (or whatever the List does, just as a 
set)


> If you put a raw HashMap into an SEI (at least, via a complex JAX-WS
> annotation I sent along yesterday), it cheerfully builds up some sort
> of Any2AnyHashMap mapping, which discards a wide variety of input with
> no comment whatsoever.

What do you mean "discards"?    I'm curious if the types that are being 
fed into the writer are known to Aegis or not.   If not, this could be a 
configuration thing to make it work, although Aegis should throw an 
exception like "Don't know how to write class BlahBlah" like the JAXB 
runtime does if you try to feed it classes it doesn't know about.    If 
it DOES know about the class, it obviously should be writing out the 
appropriate xsi:type and just plain work.

Dan


>
> Is this a good thing? Should the raw HashMap result in a diagnostic
> like the Set?



-- 
J. Daniel Kulp
Principal Engineer, IONA
dkulp@apache.org
http://www.dankulp.com/blog