Possible bugs: Web command edge handling deficiencies
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report
This file and image demonstrate a couple of different but potentially related issues with Web:
The diagonal bar running through the center is the web. The web profile is drawn on top of the base platform and extruded upwards.
First, this Web uses the "Extend Curves" option, so the far corner of the object should be solid. But in fact it has a small triangular volume missing. You can just barely see it in this image, but zoom in with the attached f3d file and you can see that it's a fully-formed gap.
Second, Web seems to be two minds about geometry on the non-extrusion side of the operation. If the selected curve goes off the base platform, Web will disallow it, instructing "ensure the profile curve can intersect the body" (well, "ensure the profile curve can intersect body" :-).
So gripe 2a is that Web unreasonably rejects profiles that could perfectly well be extruded and joined to the existing body. It isn't in fact necessary that the entire length of the curve lie directly on an existing body. If the profile curve extends over the edge, the proper treatment of that part of the curve is no more ambiguous than for the center of the curve over the base platform; there's no ceiling, so Web is inferring the proper extrusion height for the entire web from the point of contact with the far wall.
Gripe 2b is that Web obviously understands that the base platform has edges, but it still botches the handling of the edges in pretty much the same way it botches the far corner. It just seems to be blindly offsetting the profile curve and letting the chips fall where they may. If Web knows so much about the edges of the sketch plane, why doesn't it treat them identically to walls?
Third, take a more detailed look at the near end of the web, specifically, the angle of the end face relative to the direction of the web. It's not normal to the original profile curve, nor to the extended profile curve. In fact, it seems to be some Frankenstein creation formed by having Web extend the right face of the web until it hits the edge of the base platform. (Aha! So Web DOES know about the edges of the object on which the sketch is drawn!) But that's still not the correct and complete solution, because it results in Web leaving out a triangular area just as it did on the far end.
Just to be clear, here's the result I'd expect from this operation: