ART,
Just posting thoughts on the matter and don't expect ya to do anything unless it really makes to do so.
The tool table be can modified to keep original tools, copy the same tools with a different number and modified offset, etc.
no one way works for all the varity that can exist or how the machining will be done. Simpler profiles with min number of tools used
can provide a level of automation.........

RICH

ART,
How is the implementation of arc coding doing? Any progress?
RICH

Hi Rich:

 Not yet, soon hopefully..

Art

ART,
I bet you will be happy when you arrive at that point!
RICH

Hi All,
Here is a rough draft of Appendix "F". Feel free to tear it apart........
RICH

MODIFIED: NOW LOCATED IN MEMBERS DOCS

ART,
How is the finish pass progressing?

RICH

Rich:
 
   Its suprising to me how few arcs there are in the finish pass. I have found a few, and am still working on the conversion algorithm. Finding arcs from
linear segments is somewhat a difficult operation as it has tolerance issues. The techiques used conventionally are pretty bad when applied to the
linear segemnts we use. Tolerances make things a bit inaccurate and Im finding issues with cutoffs and such. Its easy to find from 3 points in the line what arc woudl fit them, but rarely is the forth point going to conform without a certain tolerance of error and that error compounds over distance and time.

  Im still unsure exactly how the arcing willwork out from the data. It may be a cas eof having to use small enough finish tolerance to get a proper
finish on those radii.

   Its the only issue im working on in Turn at the moment, as there doesnt seem to be too many reported issues other than re-arcing the output
data..

Just an update
Art

ART,
Just a few thoughts on the finish pass which would include the arc code:

Maybe tolerance shouldn't be an issue at all in some cases of the profile. After the profile is whittled down to say with-in some high point of material left, the program then just provides the profile pass with the arc coding and maybe a repeat of it / spring pass. Say there is a bump left from the finish passes, it will be removed with the arc coded pass. Depending on the material, say it was stainless, the user may deliberately want to leave some amount so it cuts thru the work hardened material.

Practicaly speaking ( only for myself ) i don't remember ever having to place a "high level tolerance" on some wierd profiled
piece. If the designer gave me a drawing on it i would probably ask him what he was smoking lately! I am thinking here something like a decorative spindle and you would have a very difficult time measuring it. Not sure any program is smart enough to account for bad / impractical design in the first place.  

Now a ball ended shaft is a different story in that the user is after something specific and it's not based on some goofy
grouping of arcs. But you do want accurate arc coding for that kind of stuff.

I guess what i am saying is that at some point after the rough / rough finish/ the program just gives the arc coded pass
and a smoothed out curve for something decorative which is relatively true. Heck that's what the person did in CAD anyway and the cad influenced it.

Maybe the above can make  your job easier. BTW, a lot of drawn profiles will require tool changes just to accomplish it unless it's a "point tool" which will work for the complete profile pass.

RICH

Hi Rich:
 The problem is that there is no Arc to build from, by the time the finish is computed, with offsets and collision avoidance, all thats left is points, or small
linear segments equalk to "tolerance" distance in length. The problem occurs when trying to get arcs from such information. There is no arc in the data
at all in that case. Useing the original arc's from any loaded profile is impossibel as it hasnt been processed for anticollision or dosnt exist.

  The cad program may have used arc's in the drawing, but there long gone by the time a finish path is created. And though intuitively youd think
you could generate arc data from thsoe linear points... it aint liek it seems..

Art

ART,
Why kill yourself trying to get that data from what's leftover?
I would have thought that you would take the dxf info and manipulate it some .
What does LazyCam do to get just the profile for example? 
I guess what i am saying is that at some point in the program you call it quits for pathing and then heres the profile cut.
Thinking here, ie; that if it took four tools to create the finish passes,  it will take four tools to do just a profile. and they will need to end their pathing at the same points to avoid collisions. So the profile ends up being broken up into segments or you just give the code for the profile in one shot and let it up to the user to place tools appropriately to cut it.

May be a good work around until you can figure out the complexities you mention.

RICH