Woot!
Cool if you could thank Leif personally. I'd really appreciate that.
Am brainstorming with Henk about bbs software. Some nice packages here
http://www.bbsdocumentary.com/software/AACPM/CPM/RBBS/ which ran with no errors and they are .com files. Pretty amazing considering they would have been compiled for a completely different machine. The .asc I think is a .bas file and the source looks fairly similar to the one Henk has (but the two bugs on his version are not there). I ran it on another board with the setup and it asked a large number of questions and then worked just fine. It even had an option to chain a program when it finishes, so it could chain itself and prevent users ever getting into CP/M if you wanted to.
I tried doing some tests with xmodem and that hasn't worked. The message coming back is "short packets". I suspect solving this will solve the dropped character issue too.
I tried 9600 and 38400 and no difference.
I also tried logging in via a terminal program on the local PC and that works fine. So the difference between it working on the local machine and logging in from a different PC (using the same terminal software Hyperterminal private edition v5.0) is the LAN network cabling, the router and the bridge software. Of these 3, I suspect the bridge software would be the first to look at. I couldn't find the source code - maybe it isn't out there, but it could be handy to trap data going through and trap an xmodem packet and see if it indeed it is short.
Re wireless links - yes the aim is to be working on a portable CP/M computer, out in the yard or anywhere in the house, and to connect to another one wirelessly, do file transfers, and then to connect to the CP/M 'internet', which in practice would be a BBS somewhere. I'd like to get data hopping smoothly between wireless nodes too. I've got a lot working there - I have a test setup with 4 boards and they are all running some software that tests for links with a 'ping', then builds up a database of all the boards each board can talk to. Then every now and then, each board will pick a friend at random and send it the list of its friends. (using a shell to xmodem). Over time, each board ends up with a bunch of text files which has the lists of what each board can talk to. I've got that working now (it takes all night to build up a good statistical database), and I've also got upgraded software automatically propogating through the network.
Next step is to take the list of each boards friends and work out optimum paths for any message. Probably just work through every combination - I think it should be a manageable factorial number if the number of boards in the mesh is small. Then put that list in another text file, and if a message arrives with a destination x, then every board knows the most efficient path to send it.
But an important part of all this is the ability to log into the internet from within CP/M. So that might require a modification to Leif's program, ie from within CP/M you send a command out to the serial port that contains an IP address, and the bridge software tries to connect to that IP address, and then sends back a message either with the connected data or with an error saying unable to connect.