[comp.protocols.tcp-ip] Need help testing FTP's REST cmd.

nelson@sun.soe.clarkson.edu (Russ Nelson) (02/08/89)

Well, I had some work that I wanted to avoid doing, so I hacked REST
and MODE B into a version of Phil Karn's NET's FTP server.  However, I
haven't yet added the capability of *using* REST into the FTP client.
Interoperability is also desirable.  So, somewhere out there is
someone who has an FTP client that understands REST, MODE B, and
marks.  Whoever you are, will you please test the REST capability of
grape.ecs.clarkson.edu [128.153.13.196]?  Grape is a MS-DOS software
repository, as well as the semi-official comp.binaries.ibm.pc archive.

Also, my anonymous friend, how did you implement the mark storage?
The only clean way I can think of to do it is to create a temporary
file whose name is derived from the 'get' filename.  This temporary
file is used to contain the marks sent out by the sending process.  If
the transfer is completed, then the temporary file gets deleted.  The
ftp client, when asked to 'get' a file whose mark file exists, will
output the latest mark using REST, and will continue the transfer.

The marks *could* be written into the user's file, but I would rather
ensure that the partial file is usable.  The user may choose to use
only what they have instead of retrying.

The marks *could* be kept inside the user's ftp process, but I would rather
make the mark storage independent of the ftp process.  After all, one of the
reasons for an interrupted transfer is because the user's machine crashed.
-- 
--russ (nelson@clutx [.bitnet | .clarkson.edu])
If you can, help others.
If you can't, at least don't hurt others--the Dalai Lama