jhutz at cmu.edu
Tue Jan 4 12:46:07 EST 2005
[Ugh; the quoting leaders are the same for jaltman's comments and mine]
On Monday, January 03, 2005 22:10:18 -0500 Sam Hartman <hartmans at mit.edu>
>>>>>> "jhutz" == Jeffrey Hutzelman <jhutz at cmu.edu> writes:
> jhutz> On Sunday, January 02, 2005 16:02:41 -0500 Sam Hartman
> jhutz> <hartmans at mit.edu> wrote:
> >>>>>>> "jaltman" == Jeffrey Altman <jaltman at columbia.edu> writes:
> jaltman> I want the hint to provide useful information when the
> jaltman> cache name is displayed to end users.
> >> I'm concerned that with the hint, this interface is too
> >> similiar to krb5_cc_resolv.
> jhutz> Because it takes the same number and type of arguments?
> jhutz> That seems a little paranoid.
> And they mean roughly the same things? I don't see how the operations
> are different except that gen_new must deal with duplicate hints.
Actually, _both_ functions have to deal with duplicates.
krb5_cc_resolv is expected to use an existing ccache if there is one.
The new function is expected to always create a new ccache. The hint is
just to help it produce a deescriptive name. This is analogous to the
difference between open+O_CREAT and mkstemp.
An alternative solution to jaltman's problem would be to give ccache's a
descriptive string, and add a new interface for getting and setting that
string. But that sounds like work.
More information about the krbdev