Shibboleth EDS and MDQ

[Peter Schober]

* Cantor, Scott <cantor.2 at osu.edu> [2020-10-27 19:15]:
> For the rest...you can't make the impossible work. There are various
> solutions to building discovery feeds out of band that the EDS can
> use, all unsupported and one-off at this point.

Well, one can always run shibd on a VM/container somewhere and let it
do the heavy lifting (updating and verifying aggregates, converting
the verified metadata to EDS JSON, then ship the file elsewhere for
local usage/distribution).

While running the SP for that purpose alone seems overkill (esp given
the SP's XML parser's limitation with very large aggrgates) so is
writing the XML to JSON conversion from scratch, IMO[1].
At least the "let the SP do it" approach is and remains fully supported. ;)

-peter

[1] Though "from scratch" may not be necessary, e.g. leifj wrote one
in XSLT for Discojuice's JSON format many years ago:
https://github.com/IdentityPython/pyFF/blob/master/src/pyff/xslt/discojson.xsl
to be later replaced with a version in Python:
https://github.com/IdentityPython/pyFF/commit/2af6de3f3d96c860032aef5f1fa93a87901429ac
Something not entirely unlike that is probably still running within
SeamlessAccess, today.