Offset command pixel dependent?

I thought I finally had it. I've been following the many threads here on
how to determine (absolutely) some point within a closed polyline. Well I
believe I found the scheme and I see no flaws -- but that's not my thought
here.
My thought is that once I developed the scheme (which I'll outline below),
the offset command may or may not work depending on the quality of the
display. Like the Boundary command it appears that the Offset command is,
indeed, dependent on the display. The closed polylines that I wished to
offset inwards at times are so small, relative to the current display zoom,
that the offset goes Outside instead of Inside. Has anyone else come to
these conclusions?
Oh, and here is my scheme for determining a point Inside a closed polyline,
irregardless of its irregular shape.

Hatch the polyline using Ansi31, to an angle of 135 (horizontal).
Explode the hatch, resulting in "loose" lines.
Extract the data from one of the lines (entget (entlast))
Calculate half the length of that line and deduce the midpoint of it.
That point is, of course, inside the closed polyline.
Erase the hatch lines since they're no longer needed.

Finally, use that point as the Offset direction pickpoint, which doesn't
always work because of display dependence. Or, is my conclusion wrong?

Len Miller

 

Len,

We can't know what's going on without seeing your code. For instance, are you using
(command "offset"... ) and if so, is osnap turned off? If the answers to those
questions are yes and no, expect unpredictable results.

An alternative: the Active X Offset method, where osnap on/off wouldn't matter.

Your method for determining point inside a pline sounds less than ideal to me. Others
have posted better methods. For instance, John Uhden's @cv_inside function.

Joe Burke

 

No, osnap is turned off. Yes, I am using (command "offset"... ). I
remember looking for a VL equivalent but my inexperience left me with an
empty find. What is it?

 

I have found generous examples on this forum for vla-offset thanks to Google
and no thanks to help within the 2005 Vlide.

I will now be looking, and I'm sure I'll find it, examples of how to make
consistent the direction of closed polylines, because their offset (inside
or outside) appears to be dependent on that.

 

Try it:

Code:

(defun is-lwpoly-clock (lwpl / pnts angl)
(setq pnts (mapcar (function cdr)
(vl-remove-if-not
(function
(lambda (a) (= (car a) 10))
)
(entget lwpl)
)
)
angl (mapcar (function angle)
(cons (last pnts) pnts)
pnts
)
)
(minusp (apply (function +)
(mapcar (function
(lambda (b)
(cond ((< (abs b) pi) b)
((minusp b) (+ (* 2 pi) b))
(t (+ (* -2 pi) b))
)
)
)
(mapcar (function -)
angl
(cons (last angl) angl)
)
)
)
)
)
(is-lwpoly-clock (car (entsel)))

 

Thanks for sharing. I will certainly look your code over to see whether I
should use my kluge or yours, which looks elegant. My kluge:
(vla-offset
(vlax-ename->vla-object (handent MinorItem))
(* OffsetValue 1) ;_inside is goal
)
(setq EntName (handent MinorItem))
(setq ObjPrior (vlax-ename->vla-object EntName)
AreaPrior (vlax-curve-getArea ObjPrior)
)
(setq ObjAfter (vlax-ename->vla-object (entlast))
AreaAfter (vlax-curve-getArea ObjAfter)
)
(if (> AreaAfter AreaPrior)
(progn
(entdel (entlast))
(vla-offset ObjPrior (* OffsetValue -1))
)
)