[bit.listserv.ibmtcp-l] VM-to-VM FTP: Corrupted module

EJR9006@UCSFVM.BITNET (Edward J. Rovera) (02/03/90)

   We are running FAL 1.2.1 on our production 4381-92E and have just
connected up our new 3090-200S to the ethernet.  Both machines are
running VM/XA 2.0 with VM/HPO 5 running underneath.  The FAL code is
running on the HPO systems, not in XA.

   In the midst of doing some FTP's between the production 4381 and
the new machine we discovered that MACLIBS were getting corrupted.
The reported symtom was that X'FF' was being converted to X'07'.
Since we are running VANILLA FAL code right off the tape (and I have
not done any futzing with any translate tables) I strongly suspect
that what happened happened due to something that came with FTP and is
due to something we didn't do during the FTP process.  Does one have
to issue a BINARY command before FTP'ing between VM systems?  If so,
why?

   I also encountered a specification exception running a module that
I had FTP'ed to the 3090 that disappears when I used the sneakernet
(i.e, VMFPLC2 DUMP to tape and sneaker it down to the new machine
room) to move the module.  I don't know if this was a random glitch or
related to the MACLIB translation problem.  Had I had more time, I
would have done more detectiving but I was in a hurry and just got
what I needed done.

   Thanks.

08071TCP@MSU.BITNET (Doug Nelson) (02/03/90)

>   In the midst of doing some FTP's between the production 4381 and
>the new machine we discovered that MACLIBS were getting corrupted.
>The reported symtom was that X'FF' was being converted to X'07'.
>Since we are running VANILLA FAL code right off the tape (and I have
>not done any futzing with any translate tables) I strongly suspect
>that what happened happened due to something that came with FTP and is
>due to something we didn't do during the FTP process.  Does one have
>to issue a BINARY command before FTP'ing between VM systems?  If so,
>why?

Yes, you have to issue a BINARY command if you're transferring binary
data.  Why?  Because the FTP spec says that ASCII is the default, that's
why, plain and simple.  In general, binary files are going to be useless
when sent from one OS to another, and there's no way to determine for
sure that both systems are running the same OS.

Doug Nelson
Michigan State University

ELINSKY@YKTVMT.BITNET (Jay Elinsky) (02/04/90)

I use EBCDIC representation and block mode (REP E, MODE B) to transfer
binary files between VM systems.  Block mode preserves record boundaries,
while stream mode might create a false end-of-record.  I think this is
in the User's Manual, but I don't have it at home so I can't check it now.

                                       Jay Elinsky
                                       IBM T.J. Watson Research Center
                                       Yorktown Heights, NY

VSBENZI@WEIZMANN.BITNET (Benzi mizrahi) (02/04/90)

On Fri, 2 Feb 90 17:59:37 EST Doug Nelson said:
>>   In the midst of doing some FTP's between the production 4381 and
>>the new machine we discovered that MACLIBS were getting corrupted.
>>The reported symtom was that X'FF' was being converted to X'07'.
>>Since we are running VANILLA FAL code right off the tape (and I have
>>not done any futzing with any translate tables) I strongly suspect
>>that what happened happened due to something that came with FTP and is
>>due to something we didn't do during the FTP process.  Does one have
>>to issue a BINARY command before FTP'ing between VM systems?  If so,
>>why?
>
>Yes, you have to issue a BINARY command if you're transferring binary
>data.  Why?  Because the FTP spec says that ASCII is the default, that's
>why, plain and simple.  In general, binary files are going to be useless
>when sent from one OS to another, and there's no way to determine for
>sure that both systems are running the same OS.
>

You can use EBCDIC command if you're transfering binary files VM-to-VM,
you may want to use MODE BLOCK  to preserve the file structure.
FAL FTP  SYstem, subcommand tells you what remote OS you're talking to.

Benzi

Weizmann Institute of Sceience,
Computing Center,
Rehovot, Israel.

SP0003@BINGVMB.BITNET (Dick McCarthy) (02/06/90)

ASCII is the default and that will warp binary files. Fig 2.2 on page
2-2 of command reference says VM to VM should be EBCDIC,BLOCK. That
should be faster than BINARY which will work too.


                                  Regards,
                                  Dick
=====================================================================
SP0003@BINGVMB.BITNET     - or -     SP0003@BINGVMB.CC.BINGHAMTON.EDU
     State University of New York at Binghamton  (607) 777-6106

KLASSEN@UVVM.BITNET (Melvin Klassen) (02/06/90)

>On Fri, 2 Feb 90 17:59:37 EST Doug Nelson said:
>>>In the midst of doing some FTP's between the production 4381 and
>>>the new machine, we discovered that MACLIBS were getting corrupted.
The directory for a MACLIB contains non-standard, non-displayable
characters.
>>>Does one have to issue a BINARY command before FTP'ing
>>>between VM systems? If so, why?
>>  ...
>>Yes, you have to issue a BINARY command if you're transferring binary
>>data, because the FTP spec says that ASCII is the default.
It's necessary, but not sufficient, to issue the BINARY command.
See Table 2-1 on page 2-2 of GC09-1204-02 'IBM TCP/IP for VM Command
Reference Manual, Release 1.2'.
Basically, it recommends to use 'TYPE EBCDIC' and 'MODE STREAM'
when transferring "text" between two VM systems.
(Experimentally, "text" can be any of the 256 hex. codes, except X'15';
although the manual states that "text" excludes X'1E'.  Hmmm.)
It recommends to use 'TYPE EBCDIC' and 'MODE BLOCK' when transmitting
"binary" files, e.g., MODULE, TXTLIB, and LOADLIB files.
In this case, I recommend to do what IBM says, and you'll be OK.

PGP@PSUVM.BITNET (Philip G. Pishioneri) (02/06/90)

Does
          EBCDIC, STREAM for readable text, and
          EBCDIC, BLOCK  for "binary"

mean that VM-to-VM (or MVS) MPUTs or MGETs of a combination of such files are
not possible?  Or can one just get away with specifying EBCDIC, BLOCK and
take a little more overhead in the FTP session for the text files?

-Phil

SP0003@BINGVMB.BITNET (Dick McCarthy) (02/09/90)

As I understand it, your second statement is true as far as reduction
of overhead. If you set type E, mode B, both binary and text would
arrive fine. If you just had text, the stream mode is less overhead
because block headers are not sent.


                                  Regards,
                                  Dick
=====================================================================
SP0003@BINGVMB.BITNET     - or -     SP0003@BINGVMB.CC.BINGHAMTON.EDU
     State University of New York at Binghamton  (607) 777-6106