Infinite loop in 2D subsystem ?

[rambo]

I ran into weird bug when designing a lasercut enclosure, all the sides of the box are made with the 2D subsystem and for visualization extruded to 3D, this works fine. But when I try to compile just one of the 2D designs alone (for exporting to DXF) it will cause a seemingly infinite loop (I ran out of swapspace at ~20G of memory used by OpenSCAD...), and it's not even the most complex part that triggers this.

This is on Mac, using 2011.12.13 version but I'm fairly sure I ran into same issue with the stable version as well (I switched to devel for a reason).

Code at https://gist.github.com/1492786

[kintel]

Thanks for the report.
This is a problem in the OpenCSG Preview (F5) - I'll look into it.
In the meantime you can use Compile and Render (F6) instead, it should work.

[donbright]

Can you do the F6 compile instead of F5 compile? On my machine F6 goes
OK, but F5 appears to be suffering from 'normalization explosion'.

-DB

On Sun, Dec 18, 2011 at 3:06 AM, Eero af Heurlin
reply@reply.github.com
wrote:

I ran into weird bug when designing a lasercut enclosure, all the sides of the box are made with the 2D subsystem and for visualization extruded to 3D, this works fine. But when I try to compile just one of the 2D designs alone (for exporting to DXF) it will cause a seemingly infinite loop (I ran out of swapspace at ~20G of memory used by OpenSCAD...), and it's not even the most complex part that triggers this.

This is on Mac, using 2011.12.13 version but I'm fairly sure I ran into same issue with the stable version as well (I switched to devel for a reason).

Code at https://gist.github.com/1492786

---

Reply to this email directly or view it on GitHub:
#44

[kintel]

Another way to work around this is to use the render() statement in strategic locations to evaluate that into a single polyhedron (thus simplifying resulting normalized CSG tree).

[rambo]

F6 compile works but since there is no way to reload without doing the "normal" compile this get's a bit complicated.

Anyway, worked around the problem by doing a 2D projection from 3D extrusion (as can be seen from the .scad file), fits my usual workflow (external editor, automatic reload & compile) better :)

[rambo]

The render() call also works fine, thanks for that tip.

[kintel]

Just one comment: The rendering is still pretty slow. That's an optimization for later.

[rambo]

render was still faster (and definitely cleaner) than making extrusion and projection.