Custom metadata provider

[Brent Putman]

On 12/9/15 7:08 PM, Cantor, Scott wrote:
> On 12/9/15, 6:20 PM, "dev on behalf of Brent Putman" <dev-bounces at shibboleth.net on behalf of putmanb at georgetown.edu> wrote:
>
>
>
>> I don't know that I've ever consciously intended to do this, but I don't necessarily object.  But gets into the whole OpenSAML api vs. impl question.
> What question specifically? I haven't intended to be crafting any mixed messages about how we're going to treat those modules.
>

I just meant that:  We've discussed that we probably can't effectively
maintain the -api vs -impl distinction in OpenSAML, as currently
defined in the versioning policy, without building a lot a
registries/factories/etc to serve as a bridge.  If we: 1) chose not to
do that (it would be a lot of work which isn't funded), and 2) don't
find another way (a middle ground) to reconcile version policy with the
codebase (i.e. tweak definitions) then 3) the other (extreme)
alternative would be to dispense in OpenSAML with the notion of -api vs
-impl entirely.

In the latter case there's no point in moving anything because
everything would be API (or we consolidate all of them to functional
modules - security, saml, etc - but thereby don't need to take into
account where specific classes need to go).

As I've mentioned, the other notion behind -api/-impl was Chad wanting
to plan for possible OSGi usage.  That didn't necessarily have the same
api/impl requirements as the current versioning policy. But that's
likely moot at this point.  I doubt for example that we're going to
re-do (again) the IdP architecture, based on OSGi.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://shibboleth.net/pipermail/dev/attachments/20151209/19ce6408/attachment-0001.html>