A while back I moved the sandbox triplesec branches to directory
trunk because I couldn't get anything to work using released versions
but could against trunk. However with the recent changes to
bootstrap schema processing I strongly suspect that everything
broke. I managed to get some of the tests to compile but I don't
trust the code I came up with. If someone (Alex) could take a look
at one of these tests I'd appreciate it.
All my recent work has been in sandbox/triplesec-jacc2; sandbox/
triplesec-jacc is primarily a monument to my newby mistakes.
The best examples are probably the tests in triplesec-store.
BTW I got annoyed at the code in admin-api and wrote something that's
sort of a object-ldap mapping framework inspired by (my poor
understanding of) jdo/jpa. It doesn't do enhancement or read
metadata from xml so you have to write java code to install that, but
the persistence stuff is then pretty automatic. I'm sure there are
parts I forgot plus plenty of bugs, but it seems to deal ok with:
-"pojos" that correspond to subcontexts/entries in ldap with single
and multivalued attributes.
-collections that correspond to children of an entry, such as the
permissions under an application. These can be separated by e.g. an
ou=permissions (or anything else that is single-valued and not
mapped). These are modeled as maps with the key being the dn component.
-collections that correspond to multivalued attributes where the
values are keys into a "child map" as just described. Again these
are modeled as maps. An example in triplesec is the grants/denials
for a role, these are multivalued attributes where the values are the
permission ids for the actual permission under the application.
To confuse the issue this code is in admin-api2 whereas the original
is in admin-api and only admin-api2 is compiled at the moment.
thanks
david jencks
