Proposal to export gssapi context

Nicolas Williams Nicolas.Williams at
Wed Mar 24 15:28:53 EST 2004

On Wed, Mar 24, 2004 at 03:14:08PM -0500, Sam Hartman wrote:
> >>>>> "Nicolas" == Nicolas Williams <Nicolas.Williams at> writes:
>     Nicolas> On Wed, Mar 24, 2004 at 02:55:29PM -0500, Sam Hartman wrote:
>     Sam> From a style standpoint, I really don't like to see unions being used
>     Sam> above the level of bit twiddling with device registers.  If others on
>     Sam> the team disagree I don't mind being outvoted on this (or other) style
>     Sam> points.
>     Nicolas> CHOICEs are discriminated unions.
> Only in the world where sequences are structs.
> Once Tom gets done yelling at you about that, I'll answer any choice-specific questions;)

I didn't mean "C unions."

> Seriously my problem is with the implementation of union in C not with
> the concept.  I agree that for the Pascal version of this call, a
> tagged varient record is appropriate.

I'd missed the fact that the proposal was to "encode" as a C structure.

> Unions are more compact by introducing the possibility of interpreting
> one data type as another without intending to do so.  Some times--for
> device registers and other low-level things this is what you want.  I
> don't think it is here.

Yeah, I agree with this.  See above.

This brings up the point: why not use a real encoding?


More information about the krbdev mailing list