Re: doi-07/interoperability questions

[Ben Rogers <[e-mail]>]

Derrell D. Piper writes:
> Ben,
> 
> >At the bakeoff, we ran into the small problem of some recent changes to
> >the DOI document which caused many machines to be un-interoperable.
> 
> The change to use an attribute to fully identify the appropriate AH transform
> occured in the Version 3 DOI, which was submitted to the ID on July 31, 1997.
> That was eight months and four drafts ago.  I'm sorry you missed it.  It's
> release might simply predate your participation on this list.
>
> I think your characterization of this change as being both unexpected and
> recent is at odds with the facts.  I also think your assertion that "very few
> vendors had actually implemented this" is grossly inaccurate as well.

I'm not complaining about the current draft.  In fact, I have
implemented it.  However, I found that sending either an AH-MD5 or an
AH-SHA1 with the corresponding HMAC-MD5 or HMAC-SHA1 attribute was
not accepted by many implementations, and only 3 or 4 others actually
sent these transform payloads with the correct auth attribute.

Perhaps I just had bad luck with the people I tried to interoperate
with.


ben