Reply 20 of 58, by wd
At the moment i can't see anything wrong with the int10/ah=10 implementation
of dosbox, it's much the same as that of the tandy1000 roms. Ralph Brown's
interrupt lists this specific interrupt call to have a different functionality on
tandy2000, that is returning pointers to font tables. The way pdm2 calls
the int10/ah=10 definitely looks like it expects this behaviour.
Now there are two possibilities: either deskmate does some detection thing
that gets it to assume dosbox is a tandy2000, and therefore uses those calls.
Or this deskmate version is some specific build for the tandy2000s or something
like that.
Mess producing garbage could hint at the second possibility, that is this pdm2
simply does not work on tandy1000. Any way it *might* be possible to just
add this tandy2000 functionality, but unless the real source of the problem is
known this is not what will be put into official sources.
How far did you get with tand-em? Does it have the same character corruption
as seen in dosbox?