-questions:
-- should do_traversal code for old-style lookups still be requesting referrals?
- If so, within what scope should they actually be used?
-- Should we do the single non-referral fallback always or only on certain
- KDC failure states? Probably answer this from testing.
-
-current:
-- rewrite verification to be more tightly-coupled to referral case
-- when should ccache be checked during referral process? never?
-- free TGTs at fallback
-- add error reporting to end of gc_from_kdc
-- deal with fetching remote TGTs after all before referrals
- - this is needed in domain_realm case
-- tgs-req realm needs to match server realm; requesting an ATHENA.MIT.EDU
- ticket, say, from NOT.MS.MIT.EDU, fails
- - rewrite initial TGS request
-- rewrite service realm before ticket goes back so that future requests
- will hit on ccache
- - testable with "kvno host/argos.mit.edu@NOT.MS.MIT.EDU"
-- write up understanding of current referral scheme to krbcore
+stuff to add:
+- write up understanding of current referral logic to krbcore
+ - given the length of conversations with hartmans and raeburn, others
+ are likely to take issue with the finer points.
+- add klist option to print actual credential principal
+- referral loop checking
+- properly return TGT string for ccache
+ - old code was convoluted and buggy. replace.
bug fixes:
-- kvno crashes freeing in_cred after the call completes. why is this?
- reproduce: "kvno host/maybe.not.ms.mit.edu@NOT.MS.MIT.EDU"
-- assertion failure: "./ptest argos.mit.edu"
- - might require NOT tickets and no domain_realm setting
+- memory management issues:
+ - kvno crashes freeing in_cred after the call completes. why is this?
+ reproduce: "kvno host/maybe.not.ms.mit.edu@NOT.MS.MIT.EDU"
+ - assertion failure: "./ptest argos.mit.edu"
+ - might require NOT tickets and no domain_realm setting
+ - no longer reproducible?
+ - fix double-free in gc_from_kdc_opt cleanup
+
+testing issues:
+- verify that cached tickets work properly
+- verify that intermediate TGTs aren't cached but
+- Should we do the single non-referral fallback always or only on certain
+ KDC failure states? Probably answer this from testing.
+- credential cacheing unreliable; investiagate
+ - "kvno host/argos.mit.edu@NOT.MS.MIT.EDU" with NOT tickets fills up ccache
low-priority:
- code (or explicitly punt) edge cases in krb5_get_cred_from_kdc_opt
-- add klist option to print actual credential principal
-- referral loop checking
later, high-priority, hard:
- padata parsing
-testing issues:
-- verify that cached tickets work properly
-- verify that intermediate TGTs aren't cached but
-
final:
- check namespace use with tom
- review code for: