On Tue, 29 May 2007 09:46:37 -0300, Shmuel Metz (Seymour J.) wrote:
>
>>Transferring variable length records as binary to
>>a stream-oriented platform is lossy.
>
>No.
>
As often, it depends on the technique:
$ ftp mvs
ftp> ascii
200 Representation type is Ascii NonPrint
ftp> get foo3
18 bytes received in 0.044 seconds (0.40 Kbytes/s)
ftp> get foo4
20 bytes received in 0.038 seconds (0.51 Kbytes/s)
ftp> quit
$ head foo?
==> foo3 <==
abcd
efgh
ijkl
==> foo4 <==
abc
def
ghi
jkl
Versus:
$ ftp mvs
ftp> binary
200 Representation type is Image
ftp> get foo3
200 Port request OK.
12 bytes received in 0.056 seconds (0.21 Kbytes/s)
ftp> get foo4
12 bytes received in 0.044 seconds (0.27 Kbytes/s)
ftp> quit
[EMAIL PROTECTED]:329$ diff -us foo?
Files foo3 and foo4 are identical
In this case, the binary transfer was lossy; the ASCII transfer was
somewhat better. I admit that there are other modes peculiar to the
mainframe that may be better yet, but safest is to flatten the PDS
to RECFM(F) with TRANSMIT or TRSMAIN, than transfer in binary.
-- gil
----------------------------------------------------------------------
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
