Page 1 of 1

Unusual bug in 'custom thru dovetail'

Posted: Thu Mar 14, 2019 12:59 pm
by lagore
I have come across an unusual bug in the custom thru dovetail, using the settings (see screenshot) I have 2 tails centered and symmetrical so therefore both left and right tailboards should be the same (as the pinboards were) but as can be seen from the other screenshots the roughing tool in the left tails operation does a strange cut on the right-hand side with an angled cut from behind the tail. This problem occurs if #1 tail width is 8mm or less but if you increase this by the smallest amount (8.000000001mm) it works correctly.
I am using version 9.0.50
Regards
Edward

Re: Unusual bug in 'custom thru dovetail'

Posted: Sat Mar 16, 2019 5:02 am
by G-Force_Admin
I'm on vacation right now, but will try to track this down and get it fixed next week.

Re: Unusual bug in 'custom thru dovetail'

Posted: Sun Mar 17, 2019 2:03 pm
by G-Force_Admin
I don't have the code with me, but I think I know what the issue is here.
During development, I usually start with one joint (left or right, and usually Y axis). Then I copy the code and modify it for the mirrored side, and X axis.
It looks like where the path deviates, I changed the X coordinate instead of the Y (and the opposite on the other axis), so it moves to the side when it should move out.
The reason the small change in value makes a difference is that there are different routines used at the edges of the boards, depending on how much material gets removed.

This should be a pretty easy fix, as soon as I get a chance.

Re: Unusual bug in 'custom thru dovetail'

Posted: Mon Mar 18, 2019 10:55 am
by lagore
No rush (enjoy the vacation), the fact that I can use a tiny increment to get over the problem gives me a workaround. I always check the generated gcode to make sure it makes sense no matter what I use to generate it even Fusion 360.
Regards
Edward

Re: Unusual bug in 'custom thru dovetail'

Posted: Sat Mar 23, 2019 4:44 pm
by G-Force_Admin
I just uploaded a new version that fixes this issue.

It was actually simpler than I thought.
I do some math to calculate the number of passes. It was combing back as 0.5, which was changed to 0 when converting to an integer number.
I've ran into this before, and just need to check that if the number of passes = 0, then change it to 1.
Adding .0000001 made the number slightly larger than 0.5, so it then converted to 1. That's why it worked with even a slight change.