Difference between revisions of "Release Meeting Minutes/2012-07-24"

Latest revision as of 17:53, 24 July 2012

Will Fiveash, Thomas Hardjono, Sam Hartman, Greg Hudson, Ben Kaduk, Nathaniel McCallum, Simo Sorce, Zhanna Tsitkov, Tom Yu

Greg

UIDs. Token parameters. UID expands to RUID. Corresponds to existing Heimdal behavior.

Simo

Make a new EUID example.

Sam

krb4 was more consistent about real vs effective UID (and dead wrong). [ also wants token for username ]

Greg

Conflicting feedback on default client keytab. Will be build-configurable.

Tom

"Secrets in /var" objection ... we already put KDC files in /var.

Simo

Secure NFS... multiple daemons. App might not care, but NFS does. So clients are more numerous than acceptors.

Sam

Default MIT ships is not that important. Linux distributions should probably agree on something.

Greg

Response sets. Won't commit current patches yet. Will do OTP commit first. (to get concrete case as an example) Nathaniel wanted marshalling functions [ for validation etc. ] [ marshalling is in key names ]

Greg

On composition side, it's easier. Not so much on the reading side.

Sam

See REST APIs.

Greg

Don't know which is best. In a higher level language, probably not a big deal. In C, it's going to be more annoying.

Simo

Needs some standards...

Sam

It might be easier to get MIT - Heimdal agreement with exploded.

Nathaniel

Need ASN.1 help.

Greg

Have some prior work that could be used. Nathaniel, we'll want fake KDC plugin for a test case.

Greg

Default keytab. Vendors likely to override.

Sam

Search lists.

Greg

Allow graceful migration.

Sam

First or last wins? Will have to revisit eventually.

Greg

Multiple relations with same name means search.

Sam

Could have a case where there are both user and system krb5.conf files, and user wants to override.

Greg

Right now, first wins.

Greg

How to specify at build time. krb5-config? Output of search list.

Tom

Nico's policy extensibility proposal.

Simo

What format for tl_data?

Greg

Depends on DB back end.