4 Oct
1999
4 Oct
'99
2:47 p.m.
Tony Duell wrote:
I seem to remember that somebody here has a GT40 that
loses the display
after a time.
I also remember that mine did this and it took me a long time to track
down the basic cause -- lost interrupts (!).
If you're referring to mine, I don't think that's the problem, because
even power-cycling the machine and restarting the program doesn't fix it.
I'm pretty sure it was a software bug in Moonlander, causing the display
list to get corrupted.
Since my GT40 ROMs has different ROMs than the standard ones listed in
the manual, and which don't seem to perform the same function, I had to
reassemble the loader to a different address. I picked 037000. It worked,
but Moonlander overwrites it. It takes so long and is so error-prone to
toggle in the nearly 256 words of the loader that I haven't yet tried it
again. I think what I really need is the RIM loader and a description of
the RIM format.
Incidentally, my 'GT40' is missing both the
standard boot ROMs and the
keyboard. I suspect I could burn the former given a copy of the code
(source or binary), but I don't have it.
Source is on my web page:
http://www.brouhaha.com/~eric/retrocomputing/dec/gt40/
If I were burning roms, I'd change the target of the two branches in the
GETSIX routine to branch back to GETSIX rather than to L.BAD. This allows
carriage returns, line feeds, etc. to occur in the middle of the download.
On the other hand, it means that a corrupted download is less likely to
signal the host for automatic error recovery. Since I was trying to download
from a terminal program on a PC that didn't know
anything about the GT40,
that was a non-issue. But it turned out that the
"ASCII download" feature
of the program would crash if the lines were too long. Probably a lazy
programmer using the C gets() function, which is a no-no.