While Loop Possible Bug(s)?

No, the debug lines were not skipped, I just happened to have cleared them before taking the screen shot.

Hakan wrote:

I tested the code in the first post on my milling machine it's running master. Can not replicate the error.
Tried with #5=25 and 26, the final circle gets done here.

I made a quick check in the code, and from what I can see it's all standard double-precision math internally.

I was testing this in 2.7.14.

Passing here accidentally, as i am away from the forum anyway, but there is a spiralroutine.
forum.linuxcnc.org/20-g-code/35812-circu...lling-routine#123743
In the meantime a bit auotdated but working and inspiring. Big +++ is the constant feed for any dia.
Only need to modify the the leadinarc and add Z levelling. The recent version is metric only, so dont post it.
Bye for the next few month...

I think this is a bug actually. Nice exercise to learn about the code.
By switching on TP_DEBUG and recompile one can see in the output that the suggested angle for the failing arc is
"radius = 1.875781250000, angle = 0.000004591468" for the 25 case (that fails), and
"radius = 1.875487290044, angle = 6.283185125288" for the 28 case (that works).
The printout is in tp/blendmath.c, but the angle is not calulcated there.
I assume this is the reason for a very small arc being made instead of a full circle. On the other hand, it could be something else.

I think you hit the nail on the head. Because if you add a p2 to the failing arc it does 1 circle.

I wonder that. It should be sqrt of x^2 + y^2 and that should always be safe.

I am thinking that there is something due to the tool compensation that is in effect. Transitions from one circle to the next and need to make a small move and somethings go wrong (just guessing),

And the format you used "G3 I- J-" is used to program full circles. How can it then do zero circles (or a very small one)?

I didn't see this issue in master so maybe someone has been working on it?

There was/is no tool compensation in effect.