In a recent note, Charles Mills said:
> Date: Mon, 13 Jun 2005 16:49:26 -0700
>
> Gil, what about the cp command on Solaris - what does it do?
>
I believe POSIX requires the "-p" option to preserve permissions and
timestamps. I haven't verified it recently; by old memory, it works
as advertised. I believe it even works on z/OS Unix.
On OS X, "man ftp" tells me:
[ ... ]
preserve Toggle preservation of modification times on retrieved files.
[ ... ]
HISTORY
The ftp command appeared in 4.2BSD.
Various features such as command line editing, context sensitive command
and file completion, dynamic progress bar, automatic fetching of files
and URLs, modification time preservation, transfer rate throttling, con-
figurable command line prompt, and other enhancements over the standard
BSD ftp were implemented in NetBSD 1.3 and later releases by Luke Mewburn
<[EMAIL PROTECTED]>.
I tried this. It works with an OS X server, and on is the default setting.
It doesn't work with a z/OS server, neither for PDS members with ISPF
timestamps, nor for HFS members. A trace shows it uses the MDTM extension,
and:
ftp> quote MDTM foo
501 command aborted - FTP server not configured for MDTM
and, in:
Title: z/OS V1R5.0 CS: IP Configuration Reference
Document Number: SC31-8776-05
# 2.8.52 "z/OS V1R5.0 CS: IP Configuration Reference" IBM Library
Server (p1 of 5)
__________________________________________________________________________________
2.8.52 EXTENSIONS statement (FTP client and server)
Use the EXTENSIONS statement to enable the FTP server to recognize FTP
commands not
described in RFC 959.
Syntax
>>__ ________________________ __________________________________________><
|_EXTENSIONS AUTH_GSSAPI_|
|_EXTENSIONS SIZE________|
|_EXTENSIONS MDTM________|
|_EXTENSIONS UTF8________|
|_EXTENSIONS REST_STREAM_|
|_EXTENSIONS AUTH_TLS____|
So, I stand considerably corrected. It's not part of RFC 959, but a
widespread convention, but not configured at my FTP site.
Trying with an OS X server (with MDTM enabled) and z/OS client, it seems
that timestamps are not preserved.
> Wouldn't the least astonishing behavior for FTP be to behave as copy and
> cp do?
>
First, I'm not defending FTP, (unlike partisans on this list who defend
the most absurd features of z/OS, based on antiquity).
Yes. It even adds to the astonishment that z/OS FTP server makes MDTM
a configuration option, and that it's supported in the server but not in
the client.
And, yes, here I'll subscribe to the argument of consistency. I've tried
hard in PMRs, with little success, to argue that the behavior of the z/OS
FTP client command:
put //DD:SYSUT1 remote-file
... for example, should exhibit as nearly as possible and locally the
behavior of IEBGENER with the same SYSUT1. But FTP client insists on
poking around in JFCBs and DSCBs, etc, and using information it finds
there to override attributes I've specified on the DD SYSUT1 statement.
Too much misguided attempt at DWIM.
-- gil
--
StorageTek
INFORMATION made POWERFUL
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
