CPNC

I'm a little late getting into this thread 'cause my eudora's not real
happy with cad_cam...

I saw Ron's software at NAMES and was really impressed. I know of
production machine shops that do nothing but CPNC with interfaces that are
staggeringly awful by comparison. I am even more impressed with Ron's
commitment to open source (GPL) the whole thing.

A student at NIST recently worked out a system for CPNC that called up Jon
E's programs on the screen, let the operator fill in the variables, figured
the g-code and put the results into an xml delimited file. This xml file
could then be saved and re-opened for editing. When the xml program was
ready to run using EMC, the CP stripped out the g-code and sent it to the
interpreter.

Since Tcl/Tk has been suggested as the language of choice here, it has some
features that are directly applicable to the CPNC system. IMO storing each
"block" of a CP as a simple whitespace delimited list on each line of a
single file is the way to go because it is so easy to recover the data into
the appropriate variables.

It would also be relatively easy to write libraries of processes that would
run the same CPNC generated program using FlashCut, MaxNC, EMC, or
whatever. Each machine's library would have process that are named the
same but which would interpret a block for its own machine.

In fact, if FlashCut proves to be obstinate about releasing the details of
its language, the details of the CPNC system and it's files would still be
available so that FlashCut could write a library that would run CPNC on
their systems. I presume that they could then sell a protected version of
that library at nominal cost.

And speaking of black boxes, for that kind of money several of us would be
happy to box up linux and EMC on a sbc, connect the parallel port pins to 5
amp stepper bricks and just not say it's linux. We could even call it
QuickerCut. ;)

Ray