I have  just realized  that some nodes  in the network  have a  default file
format of PRINT.  I thought PRINT was not  valid but it would seem  that it is
used anyway.  I have  now added  an instruction that  changes PRINT  format to
PUNCH so that LISTSERV  is able to send stuff to these  nodes. This will cause
files with lrecl > 80 to be sent in LISTSERV-punch format, of course.
 
  As I said previously I don't want  to support PRINT format because it is not
precisely defined. Under VM you can PRINT  stuff up to 132 characters long, or
133 characters if using ASA control characters, up to 250 (I think) if using a
virtual 3211  printer, and up  to *a lot* if  printing through a  virtual 3800
printer. Under other systems  it might be different, and what  is sure is that
all networking programs cannot handle PRINT  lines of the same length. I often
see files from FRORS31  being rejected by my RSCS because  they are 'too long'
(longer than what limit? I don't even know). However I might change my mind if
someone  showed me  some SOLID  proof that  lines up  to 'nnn'  characters are
accepted by all the networking softwares in the world. Note that this 'nnn' is
just another constant  like '80' and it  does not really solve  the problem of
lrecl  12,983 files  which will  still  have to  be  sent out  via some  other
protocol. I think LISTSERV-PUNCH is adequate because it is well documented and
a sample PASCAL conversion program is provided with the documentation. This is
only for nodes that can't accept Netdata format of course.
 
  Roger: why is NIHCU set to  :fformat.PRINT? I thought your node could handle
Netdata files.
 
  Eric