Network Working Group R.T. Braden
Request for Comments #368 UCLA/CCN
NIC 11015 July 21, 1972
Categories:
Obsoletes:
Updates:
COMMENTS ON
"PROPOSED REMOTE JOB ENTRY PROTOCOL"
Chuck Holland's draft proposal (RFC #360) is an excellent
document, very complete and consistent. Since the final standard RJE
protocol will be widely used on the Network, honing its definition now
will save trouble and discontent later. Therefore, I will proceed to
make a new suggestions and pick a few nits.
1. In my humble opinion, the command verb "BYE" is overly
cute; I would find "QUIT" much less offensive
2. The "(pathname)" syntax (p.5) may need some reworking.
It would be very desirable for all protocols or Network
access programs to use the same syntax for selecting a
host and socket and/or file name. (Note that the FTP
documents use the term "pathname" in the more
restricted sense of a local file system name.)
a. The PORT construction seems very undesirable,
since it depends upon a particular bit convention
of TIP's. TIP's have bent Network protocols rather
badly in the past, but surely we don't want to build
their particular socket system into an official
protocol.
b. For convenience, it may be desirable to allow hex
and octal socket numbers.
c. There will probably be other hosts besides TIP's which
will use the "(host-socket)" pathname, and some of
them may want a transmission attribute other than "T".
The proposed syntax should be changed to allow (attributes)
in (host-socket)
d. I see no reason to exclude attribute "TE", since the control
characters cr, lf, and ff exist in EBCDIC as well as ASCII.
e. There are many EBCDIC codes, and at least 2 ASCII's. The
(code) construction needs expansion.
[Page 1]
3. The syntax of OUT might reflect the fact that pathname is
required only for (disp) of "(S)".
4. It may be desirable to distinguish syntactically (job-id)
and job-file-id). For example, this would allow the command
ABORT (job-file-id)
to abort the job currently being transmitted, regardless of
its id (this assumes that multiple jobs for a given user
are sent sequentially).
5. The replies presented in the document are very good, but may
need some elaboration. For example, the syntax error messages
should be more specific. When the user enters:
OUT=(H) UCLA91: NE/ARP998.WGW.TEST
he would like the error message to indicate explicitly that
the hostname is not valid, rather than merely being told there
is something wrong with one of the parameters.
6. Experience with remote job entry to CCN via the Network has
shown that the user wants a transmission status command, to
find out how many records have been sent or received so far.
The network bandwidth corresponds in order of magnitude to
one page per second. The average output for jobs submitted
to CCN from the Network has contained about 30 pages, so
significant transmission delays are not unusual. It is important
to add a command for this purpose.
[ This RFC was put into machine readable form for entry ]
[ into the online RFC archives by BBN Corp. under the ]
[ direction of Alex McKenzie. 1/97 ]
[Page 2]
|
