RE: ftp study (technical report)

[Peter Deutsch <[e-mail]>]

[ "George D. Greenwade" writes: ]
> 
> On Tue, 29 Sep 92 11:17:28 MET, Silvano Maffeis <[e-mail]>
> posted:
> > ifi-tr-92.13                   (available electronically as ifi-92.13.ps.Z)
.  .  .
> I very sincerely hope that one of the "further results" deals with the
> portability of the files on the polled systems, especially when the file is
> (apparently) intended to be somewhat platform-independent (as I presume
> this one to be).  I'm not suggesting 8+3; however, conceptualizing ftp to
> always be to/from systems directly supporting multiple periods (as this
> file suffers from; as does the draft.part.I and draft.part.II; as do any
> number of files on anonymous ftp archives).  Once again, we return to the
> "special characters" issue!

You are under no obligation to call his file
"ifi-92.13.ps.Z" on your system, any more than you are
obliged to call the initial IAFA draft docs "draft.part.I"
and "draft.part.II" when you copy them, either. For those
systems that support multiple periods, the current naming
conventions are quite handy but if your system doesn't
support them then use whatever names you want when you
copy them across.

To be fair I agree that this makes running mirrors or
shadows more difficult on those systems that don't support
the extensions and I don't want to be too flip in dismissing
such concerns, but realistically, how many people are
trying to run archive mirrors on VMS machines? 


> IMO, there is ABSOLUTELY NO REASON to believe that files on IAFA's have to
> be actual working names on the host system (or any system!) without some
> modification on the retrieving working system.  I have seen an increasing
> number of friendly IAFA's using names such as ifi-92_13.ps_z (yes, even a
> lowercase "z"), which are much nicer to have to deal with.  Narrow minded
> Unix users will complain "but then I have to rename it ifi-92_13.ps.Z (or
> whatever) to get it to uncompress".  Big deal -- users on single period
> systems just want to GET the file (and will probably have to rename it as
> well!).

To me this is only an issue where we are recommending a
specific filename or file naming convention for files on
multiple sites. I see no reason to take anyone to task for
using the greater flexibility of their systems when they
can and I see no reason to require people to use a lowest
common denominator just because some sites would have to
rename the files when they are copied.

You would be right to complain if, as an example, the IAFA
document _required_ the use of a file naming convention
that can not be supported on multiple platforms, but this
is not the case.

Without falling back to the worst case (which I suspect is
raw MS-DOS 8+3, but there might even be worse examples of
brain damage I don't know about) I think it fair to demand
that required files have a single period, since there are
a significant number of machines with this restriction.

Still, I think it is carrying this to extremes to
criticize someone for naming a file whatever he pleases
provided it doesn't prevent you from obtaining the file
when you want it.

> This really needs to be thought about if IAFA's are really "a great need
> and challenge in the presence of drastically growing networks."

I don't dispute this, although I think you are mixing
apples and oranges here. Perhaps I misunderstand the basis
for your complaint? Exactly why do you care what people
call files on _their_ machine? Does this cause a problem
for you fetching or do you wish to set up mirrors of such
files on your machine?


				- peterd

-- 
----------------------------------------------------------------------
The constitution of the United States begins with the stirring words: 
       "We the people of the United States..."

The new Canadian constitutional accord reached in Charlottetown
and now the subject of a nation-wide referendum begins:
       "This document is a product of series of meetings..."
----------------------------------------------------------------------