Plugins and ASN.1

[Nicolas Williams]

On Mon, Oct 02, 2006 at 06:27:42PM -0400, Ken Raeburn wrote:
> On Oct 2, 2006, at 18:15, Nicolas Williams wrote:
> >1) Let the plug-ins provide their own ASN.1 DER encoding/decoding.
> 
> The duplication of code would be unfortunate, but oh well.

Yes, oh well.  The amount of code is small.  And consumers would likely
not benefit from bug fixes to a common shared library when the code's
interface requires so much of its consumers in terms of calling
conventions and is so unstable -- rather, they'd break.

> >2) Open your ASN.1 library/converntions.
> 
> I'm not sure the current interface is one we'd want opened up as a  
> supported interface going forward.  As an internal interface, subject  
> to change in future releases, and available only through  
> krb5int_accessor, I could go with, but Sam shot that down when we  
> discussed it earlier.

I'm sure that you don't want to open it.  If you had a compiler that
could generate krb5_encode/decode-style code then you could what you
suggest, maybe -- with some versioning.

> >3) Accept code into your ASN.1 library that really belongs in the
> >   plug-ins.
> 
> For the current LDAP plugin work, this is probably what's going to  
> happen.
> 
> >I think (1) and (2) is best for now.  Plug-in authors that choose (1)
> >might copy MIT's ASN.1 DER framework (avoiding symbol collisions, of
> >course).
> 
> That had crossed my mind, too....

Oops, I meant to say (1) and (3).