CLX Version 5.03 released June 12, 2000

Version 5.03 is a patch to version 5.00, 5.01 or 5.02 released 
in December 1999 and March 2000.

The new version comes with a few minor enhancements and bug 
fixes. The most important fix should be the initialisation with
AK1A nodes which has been improved at two points with the help of
SM7GVF and OZ1LQH and the fix of a long-term mail problem which
created infinite zero-length mails. However, for the latter to work,
your neighbor node must upgrade to the current version too. Otherwise,
these disturbing mails will continue to show up.


*** INSTALLING FROM SCRATCH ***

If you are a new user and install CLX for the first time, get the
clx_500.tgz file and install CLX as per instructions in the manual.
Then come back here.

The same is true for installations still running version 4.06 of CLX.
You must backup your data, install 5.00 from scratch, reload your
data and then come back here. 

Remember to run VACUUM after you have reloaded the backup data (if 
any) before you proceed.


*** UPGRADING FROM A PREVIOUS VERSION ***

If you have had CLX 5.00, 5.01 or 5.02 running, here is your way to 
upgrade without losing any data:

1. Put the upd_503.tgz archive into the /tmp directory.

2. As user clx_us shut down your current CLX software but leave 
   the postmaster running:

       $ clx -s

3. In CLX 5.00 there existed a hard link from clx_par.sample
   to clx_par and cluster_par.sample to cluster.par. As clx_us 
   remove the *.sample files, otherwise the unpacking will 
   overwrite your current config files:
       
       $ rm -f ~/config/clx_par.sample ~/config/cluster_par.sample

   If you come from version 5.02 this should already be done.

4. As root, unpack the software patch from the archive file and run
   ldconfig:

       # cd /
       # tar xvzf /tmp/upd_503.tgz
       # ldconfig

5. If you are upgrading from version 5.00 or 5.01, there is a new database 
   table which needs to be created before you can start CLX 5.03 for the 
   first time. If you were up at level 5.02 proceed to the next item.
   Here are the commands that need to be executed as clx_us:

       $ psql clx_db < ~/db/wcy_data.cl
       $ psql clx_db < ~/db/wcy_data.idx

6. There was a problem in the ml_file table where the type for f_cntr
   was originally set as int2. However, over the years its limits have
   become too small and a problem emerged at various places (PA4AB, DB0MDX)
   where CLX would not accept new mail messages any more. To find out if
   you have this problem use the following command (user clx_us):

       $ psql clx_db -c '\d ml_file' | grep f_cntr

   If the result shows int2 in the second column, you need to apply the
   following procedure. If you see int4, you are already done and you 
   can proceed with the next item.

       $ bup_db -s ml_file
       $ psql clx_db -c "drop table ml_file"
       $ psql clx_db < ~/db/ml_file.cl
       $ psql clx_db < ~/db/ml_file.idx
       $ bup_db -r ml_file

   Now do a final test: 

       $ psql clx_db -c '\d ml_file' | grep f_cntr

   You should now see int4 in the second column.

7. As user clx_us delete outdated files and directories (contest 
   data has moved to another directory in version 5.02):

       $ rm -rf ~/exec/command/contest ~/exec/command/contest_data 
       $ rm -f ~/lib/libclxccl.so.5.1* ~/lib/libclxccl.so.5.2*
       $ rm -f ~/tools/cleanup_wwv

8. Create the two executable tables Udt_QSL and Udt_MANAGER for 
   remote access using the following command:

       $ psql clx_db < ~/db/qsl_mng.cl

   Ignore the warning message about the already existing table 
   qsl_mng.

9. If all went well, startup the new CLX 5.03.

       $ clx -u



*** CHANGES ***

- Incomplete mail messages (showing up with # byte count) are now 
  deleted by running db_maint option 6b.
  {426 DK4VW}

- The \ncntr macro now is +1 to count your own node.
  {W7IUV}

- A recent CTY.DAT had more than 200 prefixes for Russia (UA). This 
  had exceeded an internal limit. The limit is now at 500.
  {RA2FA}

- Added new options 2b, 4f and 6d to the db_maint program. 2b will clean
  up duplicate WWV spots in wwv_data so you have only one entry for 
  every three hour interval). Option 4f can be run occasionally to 
  remove any duplicate entries in the home node table (us_uhn). 
  {461 DK8LV}

- New option 6d should be run once a day to sync the mailbox and forwarding 
  directories with the users' home node settings. It fixes up any mail 
  messages which are in the local mailbox and should go to a remote node, 
  {428 G0VGS}
 
- Fixed a problem where after a remote command (rcmd) the callsign of
  the user doing the command was shown in the user list of the remote 
  system after the command was long finished.
  {446 DK8LV}

- Fixed a problem in SHOW/WWV where sometimes the logger was not shown.
  {453 DK8LV}

- Eased up remote access for the SHOW/QSL and SHOW/MANAGER commands. If
  you have a partner node with a well-assorted QSL database, now you only 
  need to specify its callsign in the qsl_rr: entry of the clx_par file.

  Before this version of CLX, you had to do two things:
  a) put the callsign and the name of the remote table into the qsl_rr: entry
  b) if the remote system was a CLX system, he had to create an executable
     table for each of the commands using

         create/!udt QSL/+- show/qsl <?>
         create/!udt MANAGER/+- show/manager <?>

     if the remote node was an AK1A node, nothing needed to be done.

  Now the tables on the remote CLX system are there by default. AK1A had the
  commands available anyway.
  The remote QSL database may be further away, it must not be the direct
  neighbor (which was the case before 5.03).
  {455 OZ1LQH, 385 DL3AMI}

- "SET/LANGUAGE German" and "SET/LANGUAGE german" now do the same thing.
  {456 DK4VW}

- cty.dat file is now read automatically without having to restart CLX.
  {457 W7IUV}

- SHOW/CONTEST crashed the system several times lately because of a 
  free standing '&' character in the contest data files. Fixed now.
  {461 DL0SRB}

- The sample clx_par file comes with waz_ddd: having all zones set and 
  commented out.
  {462 DK4VW}

- Repeated generation of broken mails problem found. It happened when a 
  mail messasge was in the forwarding queue to a node but the file was
  not there. CLX started the forwarding initialization and then stopped.
  This situation should not occur - but it did in several places. We  
  suggest to run "mbx_chk -q" regularly.
  {464 DK8LV}

- Replying to a mail which had no Subject line (probably forwarded from 
  a non-CLX system because on CLX you cannot furnish a mail message without
  a subject) resulted in a crash of usr_req. Fixed.
  {465 W7IUV}

- Setting the home node now allows specifying a user and a node call
  (only when you are a privileged user). A new entry will be created if
  none existed before.
  {470 DL6RAI}

- Initialization between CLX and PacketCluster-Systems (AK1A) should now
  be working much smoother. Together with SM7GVF and OZ1LQH a lengthy
  testing session was conducted and two points were fixed:

        +-+   +---+   +-+
        |A|---|CLX|---|B|
        +-+   +---+   +-+
  Link A is connected first then B.

  * PC16s from neighbor node A are now passed through to B. 
  * A redundant PC19 for node A is now supressed when speaking to node B,
    which used to clear out the user list of A at B.
  {441 EA7WA, DL3KDV, N8UR}
  {460 OZ1LQH, SM7GVF}
  {399 DK4VW}
  {380 DK4VW}}

- Mails that are being created or in the process of forwarding cannot
  be deleted.
  {451 DG3KCE}

- New configuration command clust_loc - allows specifying nodes which
  are local to this node. Local announces are broadcasted also to local
  nodes.
  {N6WS}

- "dir oz1acb" and "DIR OZ1ACB" now work identically.
  {469 OZ1LQH}

- When a diconnect appears during BBS (W0RLI protocol) forwarding, 
  the incomplete message is retained so that in the next session, 
  it can be forwarded correctly.
  {DJ0ZY}

- Further improved dupe checking for DX spots. After WPX Contest CW
  2000, when spots were observed that were over 24 hours old, the current
  dupe checking algorithm was expanded. The shared memory block now 
  contains a time stamp for the time that a spot was received. Each time,
  a spot is not found in the shared memory, we check, if it is older than
  the oldest entry kept in the SHM or younger than the youngest, we check
  directly on the database. We compare date, time, callsign, frequency
  and the comment field (if not disabled with the dx_comm parameter). If
  it is a dupe, the spot is discarded (not forwarded, not boradcasted
  to users, not entered into shared memory).
  {DJ0ZY}

- SET/HOME_NODE for the sysop now allows specifying a user and a 
  node call. This way you can set the home node of a specific user.
  {470 DL6RAI}

- Documentation of the SET/EXIT command was not correct. Actually CLX
  knows four different exit strings by default: "/exit", ctrl-Z, 
  "***end" and "NNNN" - either must be given on a new line.
  {473 DJ0ZY}

- The cluster_par configuration parameter conn_port had never worked 
  before. It is now fixed so that you can use different ports for 
  outgoing connects.
  {474 DJ0ZY}

- "SB" is now accepted in BBS forwarding for bulletin messages.
  {472 N0PWA, N6WS}

Ben, DL6RAI