Re: EHP External Hardware Proposal

hansw wrote:

> > The problem with G-code is not the standard, but the fact that
> manufacturers of controllers have historically ignored it and taken
> advantage of the fact that there are some optional ranges of G-codes
> that were intentionally left up to their discretion. Most of them
> also add non-standard things into the "standard" words for
> "competitive" advantage.
> >

But, there is a standard set of basic G-code that almost all controls
interpret
the same way. You can do practically anything with those basic words.
I generally don't use all the exotic canned cycles anyway, because I
don't
have enough control about how they are done. And, the more complex
stuff I do is all written by some other program, that I want to be as
general
as possible.

>I wonder how the government in all it's wisdom decided to put our tax

money into writing EMC ?

I think they had a need to write a general program for controlling
robots.
Since the language was also good for machine tools, they made their code

also able to handle machine tools. Many companies find it objectionable
that a few companies that make these CNC controls charge exorbitant
amounts
to supply them initially, and practically criminal ransom to fix them
when
they break. I talked to a guy who had an Allen-Bradley 7320 (late '70s
control) that was told that it would cost about $3000 to replace the
memory
backup battery. GM, Caterpillar, Boeing, et. are getting pretty tired
of
this.

> Did the CNC industry ask them ?

Most certainly NOT! The entrenched CNC industry wants no interference
in their gravy train!

> If indeed it is our tax money that is developing a "cheaper" solution,
> will the source code be public also ?

The complete source code is on the web site, and has been for 2 years.
It is not easily available for
Windows, as the full system does not run on that platform. It is in
gzipped-tar format, which is what
makes Linux happy to read and install it. But, with the correct
utilities, you CAN read these files
on a Windows system. I have done that.

> If it is public, where is the source for the Windows NT version ?

The full system DOES NOT WORK on NT - even with a real-time extension to
NT, the performance
was not acceptable. It may be that the packages have improved since
that test was performed,
but the dicision was made about 3 years ago to go with Linux, and I
think it was totally correct.

> Maybe we have to wait the standard 30 years and then the archives will
> release it?

No, but you will either have to port it to the Windows of your choice,
by yourself, or try out
Linux.

Jon