[1816] in Kerberos_V5_Development
Re: Krb5 Ccache DLL proposal
daemon@ATHENA.MIT.EDU (Theodore Y. Ts'o)
Fri Sep 27 18:41:07 1996
Date: Fri, 27 Sep 1996 18:40:47 -0400
From: "Theodore Y. Ts'o" <tytso@MIT.EDU>
To: "Barry Jaspan" <bjaspan@MIT.EDU>
Cc: "Theodore Y. Ts'o" <tytso@MIT.EDU>, mikec@geek.grf.ov.com,
harryc@cam.ov.com, pbh@MIT.EDU, krbdev@MIT.EDU, dosdev@MIT.EDU
In-Reply-To: Barry Jaspan's message of Fri, 27 Sep 1996 13:49:32 -0400,
<9609271749.AA03404@DUN-DUN-NOODLES.MIT.EDU>
Date: Fri, 27 Sep 1996 13:49:32 -0400
From: "Barry Jaspan" <bjaspan@MIT.EDU>
- You define cc_typed_data but do not talk at all about the legal or
defined values for type. If they are all application-defined, you
should say so. However, since the addresses field is a cc_typed_data,
I doubt *all* of them are app defined.
Cc_typed_data are used as part of cc_creds, which the ccache DLL must
store and retrieve on demand; it shouldn't touch it or otherwise modify
it.
But yes, given that we're working with different implementations, we are
going to have to explode out what the definitions of krb5_address and
krb5_keyblock are. OK, I'll work on that....
- There is no explicit versioning information in the api. Is that
handled by some standard DLL mechanism? If so, will that be
translateable into Unix implementations of this api? If not, perhaps
it should be explicitly versioned.
I think we will need an explicit versioning scheme, since I don't think
windows will provide that. Thanks for suggesting that.
- There is no function to deallocate the cc_creds allocated by
cc_seq_fetch. Also, also it is implicit in the type definitions, you
should state explicitly the allocation scheme (caller allocates
structure, callee allocates contents).
Umm... yes; there was supposed to be one. Thanks for noticing the
oversight.
- cc_lock/unlock specify no-modify locks. There is different from the
normal shared/exclusive lock semantics; why?
We probably do need to revist the lock semantics. Shared/exclusive is
probably a better idea.
- What happens if you call cc_lock multiple times?
"Behavior undefined"; the caller shouldn't do that.
- Ted
