Slic3r 0.9.9 prints external perimeters slowly even when they are bridges

[darrenfreeman]

Reproducible with http://www.thingiverse.com/thing:12925 - Bridge Torture Test.

I have my perimeters set to 20 mm/s to get a quality finish, but my bridges set to 60 mm/s due to excessive sagging at lower speeds. When I try to print the above test, first it prints the perimeters either side of the bridge, at the lower speed, resulting in the extrusion sinking to the bed. Then it rapidly prints the bridge, as expected.

Even worse, I have just printed a part in which a perimeter actually forms one of the endpoints of a bridge. The expected behaviour is that it should speed up as the perimeter goes off solid material and slow down again on the other side.

The above part would have mostly succeeded if the bridge orientation had been rotated 90 degrees. The bridge is the top of a long, thin box. Slic3r decided to make the bridge along the long axis of the box, when expected behaviour is to pick the short axis.

Hopefully I can solve some of these problems in hardware (e.g. adding a fan), but this is still basically a software bug.

[darrenfreeman]

Actually the long thin box above is not quite as I described. The section with the bridge over it is roughly square. The main problem I'm having is that the perimeter which sagged is being used as an end-point for the bridge. This case really ought to trigger the bridge being rotated to run between existing solid supports rather than from one support to a perimeter hanging in space. The perimeter is in the same layer as the bridge, so it shouldn't even be counted as a support.

[alranel]

Thanks, duplicate of #657