Re: Some non-expert additions to the proposed Internet Draft

Alan Emtage <bajan@bunyip.com> Sat, 07 November 1992 19:40 UTC

Received: from ietf.nri.reston.va.us by IETF.CNRI.Reston.VA.US id aa02221; 7 Nov 92 14:40 EST
Received: from CNRI.RESTON.VA.US by IETF.CNRI.Reston.VA.US id aa02217; 7 Nov 92 14:40 EST
Received: from kona.CC.McGill.CA by CNRI.Reston.VA.US id aa11033; 7 Nov 92 14:41 EST
Received: by kona.cc.mcgill.ca (5.65a/IDA-1.4.2b/CC-Guru-2b) id AA20978 on Sat, 7 Nov 92 12:48:37 -0500
Received: from mocha.CC.McGill.CA by kona.cc.mcgill.ca with SMTP (5.65a/IDA-1.4.2b/CC-Guru-2b) id AA20974 (mail destined for /usr/lib/sendmail -odq -oi -fiafa-request iafa-out) on Sat, 7 Nov 92 12:48:32 -0500
Received: by mocha.cc.mcgill.ca (4.1/SMI-4.1) id AA18979; Sat, 7 Nov 92 12:48:31 EST
Message-Id: <9211071748.AA18979@mocha.cc.mcgill.ca>
Sender: ietf-archive-request@IETF.CNRI.Reston.VA.US
From: Alan Emtage <bajan@bunyip.com>
Date: Sat, 7 Nov 1992 12:48:30 -0500
In-Reply-To: Stephen Tihor 212 998 3052's message as of Nov 6, 18:11
X-Mailer: Mail User's Shell (7.2.3 5/22/91)
To: Stephen Tihor 212 998 3052 <TIHOR@acfcluster.nyu.edu>, eanders+@cmu.edu
Subject: Re: Some non-expert additions to the proposed Internet Draft
Cc: iafa@cc.mcgill.ca

I see that we're in danger in getting into the religious wars again
here... which I don't plan to aggravate. However, there are a couple of
things here which I believe need clarification.

One of the first things would be to point out that there is certainly no
formal connection between IAFA and archie. This is not the "IETF archie
working group" as both Peter and myself have made clear from the
beginning. IAFA was chartered to try and make the most widely used method
of information dissemination on the Internet, namely anonymous FTP, more
useful. We have made reference in the past to "archie-like tools", but
done so as a shorthand which would be widely understood.

The only place that reference to archie is made in any of the documents
is in the second ("Publishing Information on the Internet with Anonymous
FTP") IAFA document and it is only mentioned as an example (along with WWW,
Gopher, Prospero) as one of the possible services that could be
"published" from the anonymous FTP site. Nowhere has archie or any of the
other information services on the Internet been endorsed or recommended
by IAFA. In fact a substantial part of one IAFA meeting was concerned
with just what could be done in an RFC when mentioning one of these
projects. It was decided that they could be described, and I have asked
all interested parties to submit a short brief of their projects, which
after editing will be included in the final document. I have not received
any to date.

> I do not think we should recomend the use of any specific tool that omits a
> number of the anonymous FTP sites I maintain or use most often.  

And I would agree completely. However, there is and has never been any
question of IAFA recommending or endorsing _any_ of the ongoing Internet
information system projects.

Of course there is a link between IAFA and archie in the form of Peter
and myself as co-chairs, but such links are accepted and even encouraged
by the IETF since those individuals with an interest in the field are
usually the most motivated to participate in this volunteer organization.

>								 Some UNIX
> bigots assert that VMS, TOPS, MS-DOS, etc do not matter.  I am afraid that I 
> disagree.  

Again, this is a misconception which I believe is not rooted in the
facts. I have repeatedly asked for the equivalent entries for VMS, VM and
any other operating systems (to the UNIX entry currently there) in the
documents, particularly in the first one. We have received one or two
entries but they fall far short of the depth that is required: basic
configuration is only one of the criteria. Security, administration and
maintenance need also to be addressed. We would happy to receive such
entries even at this stage of the game. [One of the problems, it is
recognized, is that whereas UNIX has a fairly uniform interface, the
parameters may vary considerably on OS's for which multiple, independent
implementations exist]. The following is taken from the reported minutes
of the last (Boston) IETF:

    - Questions about the examples used in the second document were
      raised due a possible perception of "UNIX bias" in the current
      draft. The general consensus was that those individuals wishing to
      perform the exercise of constructing appropriate templates for other
      operating systems should do so and contribute them for
      incorporation into the final document. However, it was considered
      that the current draft was adequate for the purpose.


> I have made suggestions in the past that the archie project propose whatever
> modifications to the FTP spec they require to automate their function. 
> and the IAFA standardization work should help archie handle systems whose 
> directory listing format it does not grok much better.   
>
> But until one of those efforts reachs general availability allr eferences to
> archie must be qualified by the restrictions that it ignores a chunk of the
> filespace available and thus must not be treated as cannonical or even optimal.


Having made the distinction between IAFA and archie above I'll briefly
address this. 

Mr. Tihor and others have complained a number of times both directly to
us in the archie group and on USENET about the "UNIX bias" in archie
implying that we have deliberately ignored VMS and other non-UNIX
systems. As I have responded on a number of occasions, this so-called
bias was a matter of the lack of resources and nothing else. archie was
started and until very recently continued to be a pet project of lazy
grad students. I say "lazy" since the first implementation was done so
that I could cut down on my personal workload:  there was never any
thought at the time that it would become (as it has been described) one
of the "basic Internet services".  As a result, we approached the problem
at the time with the view of "least cost, maximum yield" and coming
out of a UNIX shop, and with the knowledge that by far the majority of
anonymous FTP sites were (and continue to be) on UNIX platforms, that is
where our resources went. 

A number of people have offered to donate their time to correct this
imbalance in the project.  However after looking at the then architecture
of the system we realized that such a procedure could not be done without
a complete rewrite of the system. This has now been done in the form of
archie 3.0 (a commercial product) which has just started shipping and
should be appearing at a server near you very soon.


> The process involving add request and feeding archie filenames is one whose
> documentation has never come to my attention.   Perhaps I missed the RFC or 
> perhaps its only documented informally.  In our space, the Internet world,
> I assume serious players generate at least an informational rfc or related
> document.  [Can someone point me to the document I missed?]

The description of how to request the addition UNIX archive sites to the
archie system has been widely distributed on the Internet over the past 2
years through the archie server banner messages, manual pages and help
files, several USENET newsgroups and directly from ourselves. We
receive several such requests a month. 

As far as I know, none of the operating procedures of any of the current
information systems has been codified in this way. In addition, it is
unusual for commercial system to do so... the only one that comes to mind
is the Sun NFS protocol and that is so ubiquitous as to be a de facto
standard.

I trust that this has clarified the situation.


-- 
-Alan

+1 (514) 398-8117