RE: solving the SFTP text mode issue

[Dan O'Reilly <dano%process.com@localhost>]

At 03:03 PM 5/10/2002, denis bider wrote:
> > > That is true, but I see no nice way to implement this in SFTP,
> > > because it clashes with SFTP's assumption that all files are
> > > binary. It is incompatible with SFTP's random-access concept.
> >
> > Perhaps this means that the fundamental concept was flawed...?
>
> I don't think SFTP's fundamental concept is flawed. On the contrary, I find
> it very appealing. It makes possible things that aren't at all possible with
> protocols that can only transfer a whole file at a time.

It's very obvious that the fundamental concept is based on the idea that any
file can be transferred intact and not require translation via a fixed-length
block format.  99.999% of the time, it's because the people doing the basic
conceptual design don't know anything save for UNIX and occasionally, Windows.
That's not a liability on their part (nobody knows everything about
everything), but it does translate, at least in this case, to a liability on
behalf of the concept which emerged.

So now we have (at least, some of us believe, and not just people within my
company) an expressed and rational reason why the concept is flawed, and
why in needs to be reexamined and updated.  The reaction: it's "exotic".
It's not "mainstream".  After all, the basic concept CAN'T be flawed!

> The disadvantage of SFTP's design is that it assumes files to be binary, and
> this makes the protocol less suitable for files that are not.

Exactly - and this is the reason the design needs to be enhanced.

> > But platform bias aside, the implementation of an adequate
> > mechanism seems to be only moderately difficult.  Why not
> > take a chance and make a standard a REAL standard, not just
> > one that applies to selective platforms?
>
> Sure, why not? But people like you must take initiative and/or provide
> feedback, otherwise the problem won't be solved.

It sounds, by the tone of your recent messages on this subject, that there's
no problem to be solved, as everything outside the original concept for SFTP
is an "exotic" or "outside of mainstream" system, and hence, not worth
supporting.  And along with we of the VMS world, I suspect that those in,
for instance, the mainframe world would disagree with your definition and
labelling of "exotic" and "mainstream".

And hence, my responses to your messages.  We are more than willing to work
to see this issue gets addressed; what we expect is that the chauvinism be
put on the back burner and move on to the problem at hand.  If there are
valid TECHNICAL reasons for something not to be done, so be it.  But to
label things outside of Windows or UNIX as "exotic" and "out of the
mainstream" is counter-productive to getting anything changed.  In other
words, let's work the problem rather than label things you don't see the
case for in a pejorative sense.

The goal here is not to make this a VMS-centric protocol; rather, it's
to ensure that VMS can participate fully in all transactions without
regard to the file format.  And that's simply not the case today, given the
current state of the protocol.

------
+-------------------------------+---------------------------------------+
| Dan O'Reilly                  |                                       |
| Principal Engineer            |  "Why should I care about posterity?  |
| Process Software              |   What's posterity ever done for me?" |
| http://www.process.com        |                    -- Groucho Marx    |
+-------------------------------+---------------------------------------+