Path: utzoo!utgpu!news-server.csri.toronto.edu!bonnie.concordia.ca!uunet!snorkelwacker.mit.edu!shelby!unix!hplabs!hpcc05!hpldsla!crampton
From: crampton@hpldsla.sid.hp.com (David Crampton)
Newsgroups: comp.protocols.tcp-ip.ibmpc
Subject: Re: FTP transfer of tree structures
Message-ID: <6130001@hpldsla.sid.hp.com>
Date: 31 Jan 91 16:35:26 GMT
References: <9101291607.aa05830@louie.udel.edu>
Organization: HP Scientific Instruments Division - Palo Alto, CA
Lines: 23

A year ago I blew 2 months learning FTP for a project requiring recursive
transfer; then gave up. FTP is designed to be user-interactive and provided
(then at least) no correct automatic interpretation of the wildcard. I
developed the project with rcp -r instead. That has worked well despite
the alleged poor transfer reliability of rcp. I've kept quiet about using
it so the security priests haven't gotten to me about the .rhosts file
concerns.

Since then HP Analytical group has worked with FTP to write a shell 
around ftp to allow recursive transfers with either Windows i/f or
command line. This was developed for the HP Analytical (as in chemistry)
workstations; I don't know about how generalized it may be. I am planning
to learn more about it as a possible alternative to my present process.
I'm no salesperson but would still encourage you to enquire about
HP Chemlan product. At least don't hack at ftp!

p.s. Others of my critics claim that a safe and reliable way is to
tar the dir structure; ftp the tar file; requires tar -x at receiving
end. I've done this manually occasionally; never automated it. Cheers.
--------------------------------
David Crampton
Hewlett-Packard Scientific Instruments Div Palo Alto CA
Software Mfg Engineering