- Connection script can now deal with string which are not terminated
  by a carriage return (like the "login:" in spider).

- Fixed a bug in page and buffer timeout.

- JA3BOA reported that frame were sent line by line in version 4.2. Fixed.
  Tks Hideo-san.

- Changed to nonblocking socket for outgoing telnet connection in the
  windows version. That way, dxnet won't freeze untill a telnet connection 
  is established when a script file is run. 

- Fixed a initialization sequence problem : the callsign of the adjacent
  cluster was rejected when it is already logged somewhere else due to
  a loop. Now the adjacent call is not rejected which is of course the
  correct way.

- Windows version : automatic connection with AGWpe gone at startup.
  If you want to use the AGW packet engine, you now have to tell 
  dxnet to enable it using the command :

		SET/PORT AGW[:ip[:port]]

	where the optional parameters "ip" indicates the IP address of the 
	machine AGWpe is running on (default is loopback : 127.0.0.1) and
	"port" is the port number AGWpe is listening to (default : 8000).

  To indicate in a script file that a connection must be done via
  AGW, place the keyword "agw" after the in .C row :

	.C AGW 1 F6KAL

  If you don't specify "agw", dxnet will assume by default that you 
  want to make the connection via AGW.

- Windows version : added support of PC/FlexNet32. Use the command
  SET/PORT to enable it :

		SET/PORT FLEX

  To indicate in a script that a connection must be done via 
  PC/FlexNet32, place the keyword "flx" in the .C row:

		.C FLX F5MZN-3 v F6KBF 

  Notice that flxapi32.dll must be present either in the windows\system 
  directory or in the dxnet directory.

  Tks VE4KLM.
 
- Support cluster version of 0 in PC19 protocol. Tks K2BJG.

- DX4WIN was not able to connect to dxnet via telnet. Fixed.

- DOS version : K2BJG reported some memory problems - in fact I
  discovered that the memory library I bought for the DOS version of dxnet 
  was not working as expected. This is why dxnet was crashing 
  when it was connected to a large cluster network. Unfortunatly, the
  author of that library stopped its development and no fix is avaible. So
  I decided to write my own code which can only manages XMS but it would
  use it in a better way ! BTW, as I don't use anymore the memory manager,
  the source code of the DOS version can now be available. Just tell 
  me if you want to get it.

- During a "PURG", dxnet crashed if the bin file was not existing. Fixed.
  Tks F5GVH.

- Fixed the connection time displayed in SH/TNC. It was wrong if SET/TZ was
  specified. Tks JA3BOA.
  
- Fixed some problem with Line Feed / New lines characters in the remote
  console connexion (DOS version). Tks JA3BOA.  
  
- DxNet can now support up to 500 nodes by stream (was limited to 200 before).
  Tks K2BJG.

- Done some improvements with the memory usage in MS-DOS version.

- Added domain name support for telnet connexions in the Linux version.

- The windows version says, when started for the second time : "DxNet already
  active". That can be disabled addind the switch "-i" when you run dxnet :
  "dxnet.exe -i". That way you can run more than one instance of dxnet at 
  once. Tks DF6QN.

- SH/DX <band> is now working again for frequencies lower than 148 MHz. If DxNet
  cannot detect if the user entered a band of a frequency (ie 10-meter vs 10 MHz),
  it assumes it is a band.

- ASUSER mode : AR-Cluster format is now supported. Tks OH6NK.

- DF6QN told me that a zone filter should filter both for users and
  clusters (DxNet was only check the cluster the spot were originated from).
  Done for dx and annoucement spots.

- Due to an update in the BID function, the format of a BID record changed. This
  is why the system file BID.DB moved to BIDDB.BIN (and BID.OLD moved to 
  BIDDB.OLD). You can simply delete BID.* in ~\dxnet\system\bin.
  
- DxNet now parses BID data in the mail header as well. That would limit the
  number of dupes in the mailbox. Notice that if anybody forwards a mail to the
  cluster removing all BID info (or changing it), DxNet (and, I'm afraid, any
  cluster software) won't be able to detect it as dupe. So, please take care !