Project Review: kinit -C

Sam Hartman hartmans at MIT.EDU
Wed Sep 15 11:34:52 EDT 2010


>>>>> "Nicolas" == Nicolas Williams <Nicolas.Williams at oracle.com> writes:

    Nicolas> On Wed, Sep 15, 2010 at 10:51:55AM -0400, Sam Hartman wrote:
    >> Not surprisingly it does end up being reasonably easy to build
    >> two different versions of kinit.  One gets a stub that prints an
    >> error if you use this option.  One gets the real function.
    >> Still, I'd like to wait for one of the Os distributors (either
    >> now or in the future) to come forward and say they need that
    >> before doing it.

    Nicolas> Will has already indicated that he'd rather have
    Nicolas> kinit.local...  I think he speaks for Oracle in this case.

He indicated that there was a change required for packaging and thus
building separate binaries would be desirable.  I'm asking him to do a
bit more work and actually evaluate whether that change is undesirable.
What I'm trying to say in that the complexity of having two binaries
that are basically the same trumps "Change should be avoided," but
probably not "I've thought about how much effort we've gone to to avoid
dependencies like this on other projects for 10 minutes and we've
introduced similar complexity to kinit+kinit.non-kdc to avoid
introducing dependencies in similar situations."

Now, that evaluation may only take 5 seconds rather than 10 minutes
depending on Oracle's situation.



More information about the krbdev mailing list