![background image](http://html1.mh-extra.com/html/commodore/c-net-128-v7-0/c-net-128-v7-0_user-manual_2643384010.webp)
interrupt handler with ML code to receive data from the modem. Characters
received are placed in the normal 255 byte RS-232 input buffer. This
allowed the “Main” portion of the BBS to operate with existing code with
only minor modifications to the ML program. However, it was quickly found
that the U/D protocol programs bypassed this functionality and used their
own separate code to access the modem. This is why so many of the PROTO
files had to be modified. C-Net 128 7.0 does not use an output buffer,
instead it writes output characters directly to the Swiftlink/Turbo232
transmitter for increased speed.
One of the major challenges in creating version 7 was the upload/download
protocol files and consequently the networking (which uses the Punter
protocol to send and receive files). First I had to reassemble the three
U/D Protocol files; PROTO 0, PROTO 1 and PROTO 5. PROTO 0 was the most
difficult since I did not have the source code for it. I had to reverse
engineer or disassemble the ML and then fix any assumptions the
disassembler made. Then figure out exactly what each line was doing since
there were no comments to aid me. After reassembling each PROTO file I
then tested them with the existing RS-232 modem routines. After each PROTO
was working with the existing routines my first idea was to try to “patch”
each PROTO to work with the Swiftlink/Turbo 232. I tried this for a couple
of weeks but with only moderate success, mostly due to timing problems, the
code was unstable and finally abandoned. Then I decided to make Swiftlink
only versions of each PROTO. This was a major decision because from the
start I had intended to keep compatibility with older RS-232 modems. C-Net
7 is still compatible now with the older modems with one exception; when
you set up your system you have to decide which you’ll be using since there
are now two versions of each Upload Download PROTO file (PROTO 0, PROTO 1
and PROTO 2). Getting back to the development, the U/D protocols were
ripped and rewritten. A version of each was created that would only work
with the Swiftlink/Turbo 232.
I was just about to release version 7, in fact an Alpha version was
released to a limited number of C-Net SysOps for testing, when I started
hearing how people were worried about Year2000 compliance both in the C-Net
world and DS-2. Both programs had similar Year2000 issues. Also I was
hearing how the C-Net Electronic Mail system was inferior to comparable BBS
programs. So I pulled V7 back into development and began analyzing the BBS
for Year2000 issues. The Y2K fix actually went in very quickly, probably
within a month. The first time watching the date roll over to 1/1/2000 was
pretty cool. However, I eventually uncovered a major bug in the Message
Base ML where the index would become corrupted. Now next to the UD Base
the Message Base in C-NET is probably the most complicated program.
Probably 80% of what goes in the Message Bases is done through ML. This
bug stumped me for over a month to the point where I put everything down
and walked away from it for the entire summer. Eventually I came back to
it and after tearing apart the ML and tracing every line I fixed it.
As I mentioned above the E-Mail subsystem was in dire need of enhancement
and I had actually begun work on it some months before. So with the
message base bug squished I finished the enhancements to the e-mail system
(except for a couple of things, see the Known Problems section) and about a
hundred other little changes here and there. Here’s just a few of those:
1. Reduced the size of PRG.LOGON by moving some code to PRG.LOGONTWO
(strange things happen when a PRG gets too large, believe me and read
section 21 if you plan to do any C-NET programming)