Let me explain the cause of the various problems you had:
- DMSDDL (REPLACE option: I had fixed this already but forgot to ship the new
version of LSVRECV, sorry about this. Marc's solution is the proper one, just
specify the REPLACE option.
- Error 32 from MAINE PUNCH format: due to a missing BITEARN NODESUM and a bug
in the default value. I changed a "Push chknode 'PUNCH.A YES'" to a "result
= chknode 'PUNCH.A YES'" for better speed and forgot the following statement
was "Pull . result", not "Pull result". Anyway it generated improper default
values. I am going to resend the info on generating BITEARN NODESUM and
obtaining BITEARN NODES to LSTSRV-D. I will also include a paragraph on that
!"$"!$ NODES FILE which I hope will soon be generated by LSVBITGN (but at
present I still need GENROUTS for that, obvious speed problem... :-( )
- Return code 101 from wakeup is not documented in the helpfiles, and I hate
it. LISTSERV *was* modified to purge empty files, only I assumed you got the
standard rc=4 from wakeup in this case. Oh well...
- Distribution problems: due to a philosophical problem in Send= Editor lists
which took me 10 minutes to solve. Well, at the beginning I wanted to have:
"The owner can send to a Send= Editor list IFF he is defined as owner at ALL
the peer servers" (and same for Send= Owner lists). Obviously it can't work
that way for files, because the original "from" info is lost. Files will be
handled as a commands-job in V1.5 and that should solve the problem, but at
present I changed the clause to "ANY owner/postmaster at ANY peer server can
send to a Send= Editor | Owner list". I intend to leave it that way unless
someone objects to it since it makes the checking much simpler and works
for both mail and files.
- Udo found a bug in LSVCHK which I have fixed (duplicate use of "mm" in parse
statements --> mm/dd/yy hh:mm:ss :-( )
- Ah, about PUNCH files > 80 cols: LISTSERV will automatically use a contrap-
tion of mine which I baptised "Listserv-Punch" (was not very imaginative this
day) for punching files > 80 cols. This is completely incompatible with the
Netserv GET80 format. Please listen to my arguments before killing me :-) The
reason why I did not use the GET80 format, for which conversion programs are
available on Netserv, are:
1) GET80 crashes on files > 255. Try "TELL NETSERV GET80 XR MODULE".
2) GET80 is very inefficient in some cases. Try "TELL NETSERV GET80 CHASA
EXEC (or XEDIT, don't remember which one; anyway, only one line exceeds
80 and it causes all lines to be sent as TWO cards). Besides, Listserv-
Punch does not strip records of trailing blanks if V-format (which might
be a problem for data files).
3) GET80 is not as easy to decode as Listserv-Punch. I wrote the LSVPUNCH
program in 30 minutes in assembler (and didn't enjoy it :-( ), so I
guess it would not take much time to write a decoding routine in ANY
decent language. The Listserv-Punch format is explained in the LISTPUN
MEMO which is not yet available ;-)
- Why did I get to the pain of creating a GET80-like format? Well, I hope to
add file server capabilities to LISTSERV as soon as possible. First, we need
it for NOTEBOOKs created by the "Notebook=" keyword. Second, it would be a
GREAT help for lists like RSCSMODS where the mods could ACTUALLY be made
available on the servers, and restricted to the members of the RSCSMODS list,
for example. Besides, the distribution of the files would be very efficient.
Please note that I do NOT try to change LISTSERV into a Netserv-like thing.
Netserv is a very good tool and I see no reason for LISTSERV to become a
concurrent/rival server.
All of this will be called version 1.4a, and I'm sorry for all the bugs :-( I
knew I should have waited and tested it thoroughly before shipping it, but I
just got fed-up with the 20 pieces of mail saying "when will 1.4 be ready?" I
got in my reader every so often. Will teach ya! :-)
Eric
|
