krb5_get_in_tkt_with_keytab no longer honors creds.endtime?

[John Hascall]

> >>>>> "John" == John Hascall <john at iastate.edu> writes:
>     John> Somewhere between 1.2.6 and krb5-1.3.2-beta3
>     John> krb5_get_in_tkt_with_keytab() seems to have changed so that
>     John> setting creds.endtime (as documented) has no effect of the
>     John> life of the credentials acquired.
> 
>     John> Is this a new "feature"?  (it is quite annoying)
> 
>     John> It seems to me like, after krb5_get_in_tkt_with_keytab()
>     John> calls krb5int_populate_gic_opt() it ought to do something
>     John> like this:
> 
> Wouldn't it be better to pass creds into krb5int_populate_gic_opt and
> do the work there?

    That seems reasonable (I am assuming krb5int_populate_gic_opt isn't
    called a million places), but Tom Yu's message suggests it might be
    deprecated...
> 
> I'd certainly be happy to evaluate a patch that did this for 1.4 or
> 1.3.4 if that ever happens.

    OK.

John