Password sync plugin, and questions about plugin criticality

Jeffrey Hutzelman jhutz at cmu.edu
Tue Jun 27 18:34:18 EDT 2006



On Tuesday, June 27, 2006 02:53:51 PM -0500 Nicolas Williams 
<Nicolas.Williams at sun.com> wrote:

> On Tue, Jun 27, 2006 at 02:14:12PM -0400, Sam Hartman wrote:
>> 1) Always take a krb5_context.
>>
>> 2) Always take it as the first argument.
>>
>> That's a basic style guideline for almost all of our code.
>
> All the other krb5 plug-in SPIs certainly need this since they provide
> krb5-specific services to the framework/caller.  Requiring a
> krb5_context as a matter of style seems just fine.
>
> My [minor] concern was that, and I'm talking about this facility only,
> an SPI that is bound to the krb5 ABI would disallow binary compatibility
> for password sync plug-ins across caller implementations.

I don't think that's an issue, as long as it's understood that the 
krb5_context is only usable for making callbacks into the same caller 
implementation. 



More information about the krbdev mailing list