[LispM-Hackers] Serial port progress
John Morrison
jm@mak.com
Sun Mar 31 04:47:01 2002
Hi;
"James A. Crippen" wrote:
> It seems like you're making lots of progress... Are you going to be
I was going to try and respond with some humorous response about how
proud I am to be able, in the Year Of Our Lord 2002, to be able to print
ASCII characters at 9600 baud with what used to be a supercomputer (in
terms of performance) not so long ago. However, all my attempts came
out as sarcastic, so I bailed.
> adding your work to the CVS repo any time soon? I'd say just create a
> subdirectory and put all the asm, etc in there. Don't need a fancy
> makefile and everything yet, but it'd be nice to have it stored
> centrally.
Yes, I should like to. I currently have a kind of bizarre directory
structure set up, and that's why I haven't committed it yet.
root
e3-stuff (explorer simulation stuff)
common (the OS-independent infrastructure)
arch (build targets are rooted here
i386 (bare iron)
host (should really be linux/freebsd/solaris, etc)
Also, in order to support many concurrent targets in the same directory
tree (not necessarily a good idea), all object files (.o, usually)
corresponding to C++ files (wherever they are found) live in the
arch/target directory. That way, a given source file in either common
or e3-stuff can have many .o files corresponding to it, none of which
live in either common or e3-stuff. Makes the makefiles a little
messier, but I find I work in mini cycles -- making a change to the
Linux build, testing it, and then rebuilding it for the i386. Because I
do so many mini-cycles (many per hour, unless I'm well and truly stuck),
this is very convenient.
However, maybe I can make the common directory a subdirectory of the e3
root directory, and then you all can ignore what I'm doing, and I can
reference e3 root directory files from the subdirectories, and make them
in the arch/i386 place.
Does this sound like a plan? If so, I can commit the directories later
today.
-jm
p.s., please excuse the messy i386 stuff -- there's a lot of magical I/O
port stuff constants etc sprinkled around, and a lot of exploratory
branches ("how come this bank select stuff ain't working?????") hanging
about. I tend to clean it up incrementally after I figure out what the
issues are and it's working so I don't break it. Consequently, it
appears messy.
--
==== John Morrison
==== MAK Technologies Inc.
==== 185 Alewife Brook Parkway, Cambridge, MA 02138
==== http://www.mak.com/
==== vox:617-876-8085 x115
==== fax:617-876-9208
==== jm@mak.com