[ih] Fwd: Fw: Quantifying OSI

[Bob Purvy]

My own experience with X.400:

I'm chagrined to note that I was at 3Com, still believing it was The Future
in 1989. So was Eric Benhamou, our division VP (at the time). He hired
Telesystemes Reseaux in Paris to work on it for us. We sent them detailed
specs, and I even went to Paris once. Jim White, was the Special Rapporteur
for it, hired me in 1985, then left to start a consulting firm for it.

The project eventually failed, with mutual finger-pointing. It was an
insanely complicated spec, as you said, and the amazing thing was that it
remained The Future as long as it did.

On Mon, May 11, 2026 at 2:00 PM Karl Auerbach <karl at iwl.com> wrote:

> On 5/11/26 9:07 AM, Bob Purvy via Internet-history wrote:
>
> > How many hours did OSI proponents spend actually writing code and getting
> > networks running? Writing papers and going to meetings doesn't count.
>
> I spent far too many hours trying to implement X.400 (beginning with
> ASN.1/BER).  Much of that time involved pounding on desks and screaming
> "What the F!!! does this text - xxxx, yyyy, zzzz - mean?  Why are they
> doing this? This has no limits!! Will anyone ever use this part? !!!!!!"
>
> The implementation was no harder than doing something hard - like IPv4
> reassembly of fragments [I've got some test code that will knock out -
> sometimes even crash - pretty much any present day instance of IPv4
> reassembly - fortunately most of those cases, although RFC legit,
> essentially never occur in practice - but they could if sent by an
> hostile peer or done by a hostile man-in-the-middle.]
>
> But many more hours were spent trying to figure out what the standards
> meant.  And there was really no one to ask for clarifications or
> insights.  And there were few peer implementations one could test against.
>
> Overall, implementing OSI stuff was at least an order of magnitude
> (probably more than a single order) more time consuming than for TCP/IP.
>
> At least I was able to recoup some of that time investment when I did
> SNMP clients and servers.
>
>  From my implementer's point of view, ISO/OSI pointed an automatic
> weapon at its own feet and opened fire.  There were good ides in ISO/OSI
> but their proponents buried them under mountains of dross verbiage, nary
> a paragraph of which bothered to explain they why's and how's.  The
> death of ISO/OSI was essentially self inflicted.
>
>          --karl--
>
>
>