In February I fixed this bug:
https://github.com/LinuxCNC/linuxcnc/issues/1528

But, it seems, I failed to properly consider full-circle spiral arcs, and
the test was wrong.

https://github.com/LinuxCNC/linuxcnc/issues/2169
(Note that this issue report was _not_ raised by the original reporter, and
furthermore seems to show a _different_ issue.

I investigated and found a test case that I could reproduce, and put in a
fix for that. (linked in the issue report above)

However it is reported that the original bug (that I have never reproduced)
still exists.

Here is the problem description:

"if open axis.

Comand g2 x0 y0 j5 z-2 f1000.

The machine not execution g2 , but only Z-2."

I think that the system needs to be set up for immediate homing to 0,0,0 as
I have an unconfirmed report that this specific problem does not occur if
the axes have moved.
(But then the test cases that Tom_L shows in the bug report has the third
arc going wrong too)

The problem that I have is that my system appears to be working perfectly,
so I can't investigate the issue here.

I could revert the original patch, but then that reintroduces a known bug,
so I don't want to do that.

So, first help needed, someone who can compile from source _and_ can
reproduce the reported problems. (think that there might be two, and one
might well be an uninitialised "start" for the arc in the just-homed case)

This clearly can't go out as-is, I am only glad that I didn't put the
original fix in 2.8 (I thought that I had, actually)

-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is designed
for the especial use of mechanical geniuses, daredevils and lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1912

_______________________________________________
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers

Reply via email to