Happy Talk by R. O’Farrell
One of the awkward things about CP/M is that it is exceedingly difficult
to get programs and data from one machine to another, unless they can read
each others discs, like the Gemini M-F-B system. The patches to the BIOS to
allow multiple format disc read/write are not easy. One method is a modem
transfer program (BSTAM or MODEM 7). If one doesn’t have one of these, then
usually one resorts to PIP.
One can connect up the serial port on two dissimilar machines, ensuring
that the same baud rates, parity and so forth are selected, and using STAT
to change PUN: to be PTP: (paper tape punch) on one and RDR: to be PTR: (paper
tape reader) on the other. One connects the two serial ports together with an
appropriate cable, checking that the transmit and receive lines on the
different machines are connected to their opposites on the other machine. So
on the reader machine one enters:
and on the transmit machine:
For ASCII files, such as assembler listings, this works 0.K. up to a length of
about 26k, depending on CP/M size. Intoxicated by power, we then try a COM
file. It may appear to work. But STAT or XDIR or SD the received file, and
compare the filesize with the untransmitted version. Diasaster has struck!
Most probably, the received file is dramatically shorter than the original.
What has happened is that PIP, on receiving a CTRLZ (1AH) from the serial port
has interpreted that as an end of file character. In the case of an ASCII file
it would be, as an ASCII file contains only characters 20H to 7FH, with CRs
and LFs. In the case of a COM file, the 1AH is an instruction, meaning LD
rape – most useful! The specification of PIP suggests that selecting option
[o for PIP should cause it to ignore ignore the normal CP/M end of file mark,
and look to the Console (keyboard) on the receiving machine for the EOF. If
this is properly implemented in BDOS, and it usually isn’t that I know of,
then when all disc accesses on the transmitter have ceased and the cursor is
restored, typing CTRLZ on the keyboard of the receiver will cause it to break
into a flurry of activity and save the file to disc. Very rarely have I found
this to work – why, I don’t know.
Another method of file transfer is as follows: Using a disassembler, take
the file you wish to transfer (FILE.COM). Disassemble the first 4k of it
(say). When you reach the appropriate part of the Disassembler input, tell it
that the entire 4k block is DEFB instructions. i.e., treat it as a data area.
This avoids any need to fiddle around trying to sort out labels. Take the
FILE.ASM produced by the disassembler, and assemble it, linking it if
necessary, to produce FILE.HEX. Then PIP this from machine to machine. There
is no need to select any options, as being a HEX file, it is ASCII. On the
receiver, just LOAD it or ZSID it (can use DDT instead) to convert back into a
COM file. If the COM file you start off with is longer than 4k, then do it as
a number of 4k blocks (possibly 6k, depends on your machine and CP/M).
Remember that a HEX file produced from a COM file multiplies its length by
approx 2.65. So the 4k block becomes about 11k. It is better to be shorter
than longer, as with disimmilar machines the buffer lengths can differ
dramatically. It is disheartening to hear one machine start up its disc to
save the file on buffer full while the other machine is still transmitting.