The problem is that LISTSERV has to  know where to get the keywords from.
This is accomplished through the use of database "drivers", which process
the various  files and  create an index  with the  necessary information.
There is a built-in driver for  mailing lists, which knows how to extract
date, subject  and author from  a RFC822  message. It is  highly unlikely
that  any of  the built-in  drivers will  understand the  format of  your
questionnaires. More info is enclosed below.
 
  Eric
 
>>> Item number 6203, dated 92/02/07 19:56:27 -- ALL
Date:         Fri, 7 Feb 1992 19:56:27 +0100
Reply-To:     "Forum on LISTSERV release 1.7"
Sender:       "Forum on LISTSERV release 1.7"
From:         Eric Thomas <[log in to unmask]>
Subject:      Re: Listserv Searchable Databases
In-Reply-To:  Message of Fri,
              7 Feb 1992 11:14:06 CST from "Forum on LISTSERV release 1.7"
              <LSTSRV-L@SEARN>
 
On Fri, 7 Feb 1992 11:14:06 CST Marty Hoag <NU021172@NDSUVM1> said:
 
>   There isn't  any doc that  I know of except  for the REXX  code. Eric
>offered me something in French but since I can't read French I passed on
>that.
 
I actually found out that the EARN  people wrote a memo about this. It is
available via 'INFO  DBCREATE' from any LISTEARN. I have  not had time to
read it  thoroughly, but the information  seems to be based  on the notes
Marty mentioned, so  it should be accurate. There was  no major change in
that  aspect of  LISTSERV  between 1.5o  and 1.7.  There  is little  that
someone familiar with REXX can't figure  out in say 30 minutes by reading
one of  the standard  drivers (LSVDBPN for  instance). Most  drivers were
written using cut+paste from this driver or LSVDBNB.
 
Also, Silvia Giordano wrote a 'driver generator' that lets you define the
database in formal terms (with a grammar); her program will then create a
REXX driver for you. There are restrictions on the type and format of the
documents it supports (it was designed  for EARN documents and may or may
not  be suitable  for other  things). The  manual is  available via  INFO
DBDGEN from  any LISTEARN, and contains  introductory information similar
to the DBCREATE memo.
 
But as  Marty said,  if you  are familiar  with REXX  you should  have no
trouble using the  "ultimate documentation". I have wasted  the last hour
reading RSCS  V3 exit customization manuals  in the hope that  I wouldn't
have to learn about the new organization of RSCS V3 internals. Well after
1h I hadn't  found where the local nodeid was  stored, what *exactly* you
get from RSCS at exit 19, and where RSCS keeps its command table so I can
scan  it  and  find  which  ones   are  privileged  (I  wanted  to  allow
non-privileged commands from non-trusted  sources, but this is apparently
not feasible without  a mod). I found  the answer to all  questions in 10
minutes by reading DMTCMX.
 
  Eric