Markku,
>
> If you consider dynamically loadable modules (DLL's) kernel
> extensions, then you are right. It should be possible to add support
> for a totally new extension header to an existing stack by a loadable
> module, which attaches to the stack, without touching the stack and
> API code itself.
>
> This requirement does have some effect on API, and this is why I have
> some concern that the current "hardcoded known headers approach" will
> differ too much from what I have to put in.
>
If dynamic loaded modules are available to extend the stack then it is
straight forward to use them to extend the existing API.
This is largely an academic discussion, however. To my knowledge there
are no current proposals for adding new extension headers and as far as I
remember there have never been any such proposals.
If an implementor wants to code their implementation for an event that
isn't likely to ever come that is certainly the implementors choice.
However, I would object to existing code of other implementors being
discarded wholesale to make it easier for an implementor to make that
choice.
Tim Hartrick
Mentat Inc.
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page: http://playground.sun.com/ipng
FTP archive: ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------
