using Velocity to externalize configuration properties ?

[Brent Putman]

On 8/1/13 10:41 PM, Tom Zeller wrote:
> Sure. I know I have asked before, but I'll need to be reminded
> regarding the details of the incorrect v2 behavior.


Below is what I said on the committers list back in Feb.  Basically, if
we've gotten to the point of defining /META-INF/spring.schemas files, we
need to make sure we're using them correctly.  The key of those
properties files should be the (remote) schema location URI, NOT the
schema namespace URI.



> I was skimming this to refresh my memory and remembered something that
> seems relevant to mention:
>
> Re: section 'F5.2 META-INF/spring.schemas', the v2 code flubbed this.  I
> pointed out to Chad a long time ago (when I was basically learning this
> stuff the first time), but basically it was too late to fix.
>
> That file is suppose to define a mapping from a (likely remote) schema
> *location* to its local classpath location, so can be resolved locally
> rather than remotely. (It's basically their proprietary notion of an XML
> Catalog type of thingy).  Instead, our files in shib-common and
> shib-idp2 are defined as a mapping from the schema namespace URI to the
> classpath location. (The mistake probably arises from conceptually
> equating it to the schema.handlers, which is a mapping from namespace
> URI to handler class).
>
> However, despite the mistake it "works" IIRC because we also supply a
> custom entity resolver to Spring that makes the same conceptual mistake
> (at least we're consistent).  And I *think* that only works b/c we only
> ever process our own custom schema with that particular code.  If we
> were ever having Spring process both custom and standard bean schema in
> the same go, this might actually break things.
>
> Just wanted to point out in case anyone looks at the v2 stuff and says
> WTF...  



-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://shibboleth.net/pipermail/dev/attachments/20130802/7a94a24b/attachment.html