Re: Re: Re: the open source 'thing'

Chris L

There is really nothing all that difficult about the open thing. It only
means that the "man readable if you know how" code is available along with
the runnable binaries. There are many variations of open ranging from
artistic licenses through the copyleft stuff from the Free Software
Foundation, to public domain.

Much of the EMC is public domain and anyone can do anything they want with
it. My impressions are that portions of the EMC code have found their way
into several other hobby level CNC programs and may be finding their way into
some European controls as well.

A second very important aspect of the "open" movement is that people strive
to develop documents that also can be had and read in similar fashion.

Now what you do with that open code is your own problem. If you don't want
it, don't take it. If a BDI shows up on your doorstep, or as an insert in
your favorite magazine, mutilate it immediately and throw it away. If the
problems reported on this list serve to reinforce your dislike of the EMC, it
would be no different than reactions to the many CAD versions and problems
discussed in these parts. I doubt that it matters one whit to any of the EMC
folk that you buy in by taking the time to set it up. They'd be happy to
help if you do.

Several of the EMC users have been accused of being evangelists so you may
have to do the same with us as you do with other evangelists. I doubt that
you can alienate us enough so that we all get off list and leave only
MS-Windows believers -- others much less civil that you have tried. <g>

(other comments mixed in)

>    From: Chris L <datac@...>

<s>

> I realize that FlashCut is not open because of the unknown language sent to
> the Box. Beyond that, we do not know what is programmed into the main chip.
> I assume then, that  neither would be CNCPro, Master5, KellyCam, or DesKams
> product because it is a "canned" program. Unless of course, they give you
> the ability to disassemble, modify, then recompile the program either
> freely or by purchase. That is correct, Right? With the ability to
> re-compile you then could make any change you would want to the program.

Yes but the ability to recompile is not the same thing as "open source." It
might be easily possible for someone to release some of their code and still
require that you use their binary library whenever you want to recompile. In
fact many of the MS software writing tools work exactly this way.

> Now with Constant Contouring always turning up as a real tricky issue to
> get to work flawlessly, Do you feel that recompiling any of the above would
> bring it into that "sweet" spot of excellent Contouring ? Or, would we be
> saddled with base code issues that would still, even after having full
> access to the code, be limited in the contouring area ?
> I am assuming the latter.

The EMC uses cubic splines to compute changes from one motion command to the
next. According to the fellows who wrote the EMC countouring routines;

"First-order interpolation, in which the in-between points fall
proportionally between the end points. At the transition, position is
continuous but there is a jump in speed, and acceleration has large
spikes.

"Second-order (quadratic) interpolation, in which the in-between points
fall on a curve between the end points. At the transition, position is
continuous, and at one side (just preceding or just following the
transition), speed can be continuous. Since there's no point in only
having speed continuity half of the time, second-order interpolation is
not of practical use.

"Third-order (cubic) interpolation, in which the in-between points fall
on a curve between the end points. At the transition, position and speed
are continuous, but there is a jump in acceleration, and jerk (the
change in acceleration per change in time) has large spikes.

"Fourth-order (quartic) interpolation, in which the in-between points
fall on a curve between the end points. At the transition, position and
speed are continuous, and at one side (just preceding or just following
the transition), acceleration can be continuous. Since there's no point
in only having acceleration continuity half of the time, fourth-order
interpolation (like second-order interpolation) is not of practical use.
Neither are any other even-order interpolators.

"Quintic interpolation, in which the in-between points fall on a curve
between the end points, like the cubic. At the transition, position,
speed, and acceleration are continuous. Jerk is bounded and doesn't have
spikes. This is the smoothest..."
source www.linuxcnc.org/handbook/part3/interpolation.html

The fact that the EMC code has not reached the pinacle of interpolation has
bothered some users enough to ask when this might happen and whether they can
help with it. It hasn't stopped many users from doing some rather complex
contouring with what is already in there.

> So, the matter of "changing" things, in order to make this all worth while,
> really needs a "root" program that is an *awesome* program that  works
> flawlessly, Contours like crazy, has all the I/O anyone could ever want and
> they GIVE you all of the source code so you can change it.
>
> I'd like to think that when someone gets THAT program developed, one will
> not need to change it!

Whoa! "Right turn, Clyde." All you'd have to do is watch 10 machinists make
the same part to realize that there are ten ways to make a part. The same
truth applies to the machines they run or the machines made by members of
this list. It really doesn't matter how good Mazatrol is at computing
feedrates, someone is going to go through the program and change the values.
They would also change display screens and what button does what if they
could.

You can get close to what you define as these killer "programs" from vendors
like Mitsubishi, Fanuc, Fagor, Mazak, Siemens, etc. They just happen to come
with some hardware and a significant price tag. And you don't happen to be
able to look at the code from which they were compiled or the maths that they
use to compute interpolation.

> The whole topic almost reveals that to date, a really, really good PC
> windows *or* even Dos controller has not yet been made... otherwise
> everyone would have a target project to find out how it works, copy it
> without ending up in court, then give it away in a fashion that anyone can
> modify it. Not sure how many are lining up for that project.
>
> So, your right. Maybe I just don't see the issues about just "changing"
> things because I can. I could be too old for that. (Might be why my Astro
> Van does not have that Big Block in it yet.)

Premonitions of my rapidly approaching demise was what stopped me from
running 365 hp 327 Corvette engines in war surplus jeeps. It doesn't prevent
me from adjusting the seat of my car to be more comfortable for me or
changing to heavy duty shocks and brakes. I do what I can.

> The way I see it, we need a really good Windows Control. It has not been
> introduced yet. When someone finally does go thru all the effort and
> expense, I don't think they will give it away or let you modify it. If it
> is really good, I'll just use it the way it was made.

And here we need a couple small language clarifications. Windows is not a
trademarked term -- MS Windows may be. Windows95 and such certainly are. The
EMC can and does run under windows -- X-Windows. It doesn't have to but it
can. And that same X-Windows system can run decent file or web browsers like
netscape, mozilla, konquorer, nautilus, opera, or amaya. For other graphical
aps, and all of the wonderful MS/Intel stuff like USB and printing, we muddle
through.

X-windows is not a copy or look alike of any of the MS products. It was an
independent effort to create a graphical presentation environment for unix
OS's. I saw examples of it in working networks back when I first saw MS
Windows3.0. I even saw it turn a vt100 text terminal into a graphical
desktop.

In light of this, I take it that what you are saying in this paragraph is
that no one has made an perfect control for MS-Windows. Having for the most
part abandoned Microsoft, I know little to nothing about usable CNC under
these operating systems except for watching the anguish of a couple of
reasonably high level Mitsubishi engineers trying to shut down MS-Windows so
that it would connect correctly to their proprietary servo motion control
board the next time it was started. I believe that they were wishing for a
scripting macro like the old Dos .bat files. Maybe these are available now.

Since I have at least one humanities degree, I need to ask, "uh, do you want
servos or steppers with that lathe program.? Or was it a mill?"

RayH
Michigan's U.P.