DynamicHTTPMetadataProvider

[Brent Putman]

On 9/20/16 1:29 PM, Tom Scavo wrote:
>
> I'm glad to hear V3.3 is a likely target. Is this firm enough so that
> you feel comfortable updating the JIRA ticket to that effect?

Maybe. I was planning on working on that this week, since I expect that
it will be a small amount of work (other than the custom Spring wiring
in the IdP. That's always a pain.)

>
> I think it's okay to default to filenames based on the SHA-1 hash but
> I hear that the Shibboleth MDA will default to filenames based on the
> percent-encoded entityID. Is there synergy in coordinating the
> defaults across projects?

I hadn't actually heard that.  Was this coming from Ian?  Maybe it's
been mentioned somewhere and I just missed it.

If that's true, then we should certainly discuss.  We recently had some
quick comments around it on the dev call and the "consensus" there
(i.e. of anyone who spoke up) was that anything other than a (probably
hex encoded) digest of the SHA-1 hash was likely to not work reliably
on all platforms, like Windows.  If there is a more "human-friendly"
form of the entityID that we can guarantee works everywhere, then I'm
not opposed.

Not being a Windows person, I'm not personally in a position to
evaluate naming schemes there that aren't simple alphanumeric.

I suppose one way to support both schemes (or indeed arbitrarily many)
would be: Instead of having just one strategy, have multiple ones, in
some preference order.  When resolving, loop over them in order until
you find a match.  I'll give that some thought also. Assuming it's
feasible, that doesn't actually address though what the default one(s)
should be.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://shibboleth.net/pipermail/dev/attachments/20160920/2a13689f/attachment.html>