[PD] Support for point sections in loft

[AjinkyaDahale]

The BRepOffsetAPI_ThruSections method that forms the crux of loft supports the first and the last sections to be vertices instead of wires.

Addresses issue#4607.

[donovaly]

What is the status of this PR?

Concerning the feature, it should support to use vertices of 3D objects as well as sketches just consisting of a point object but not datum points.
The sketch point support might be tricky since I remember a discussion once I proposed to change the Hole feature so that one only has to provide a point. At the moment one has to produce a circle, but only the center pint of it is used as position of the hole).
So if it is possible to have only a sketcher point, I would try if I can change the Hole feature according to your solution for lofts.
(Maybe this is not that tricky, I simply never looked into this yet.)

However, I could not resist to merge already the 2 typos you corrected in the PR.

[AjinkyaDahale]

What is the status of this PR?

I answered some questions in #5168 (comment). In summary, we can use vertices of the solid object as the last section for a loft, and I'm working on trying to support solid vertices for start and later sketch vertices.

...The sketch point support might be tricky since I remember a discussion once I proposed to change the Hole feature so that one only has to provide a point. At the moment one has to produce a circle, but only the center pint of it is used as position of the hole). So if it is possible to have only a sketcher point, I would try if I can change the Hole feature according to your solution for lofts. (Maybe this is not that tricky, I simply never looked into this yet.)

If anything this support forces us to address supporting vertex-only sketches, or only vertices from sketches, because for loft (and pipe) vertices and circles mean different things.

However, I could not resist to merge already the 2 typos you corrected in the PR.

That's perfectly fine :)

[AjinkyaDahale]

Started a forum thread to keep the PR thread clean: https://forum.freecadweb.org/viewtopic.php?f=19&t=63697. @donovaly @adrianinsaval perhaps we should move the discussion there.

[AjinkyaDahale]

@donovaly @adrianinsaval there are a few things remaining to be done, but overall I would call this ready for review.

[freecadci]

for feature branch PR_5170. Pipeline 408635926 was triggered at 08e6c44. All CI branches and pipelines.

[freecadci]

for feature branch PR_5170. Pipeline 409255305 was triggered at 557193c. All CI branches and pipelines.

[freecadci]

for feature branch PR_5170. Pipeline 411472361 was triggered at f9a50d8. All CI branches and pipelines.

[AjinkyaDahale]

Adding back the "do not merge" tag, because after rebasing with PR #5176 merged, the "front" face is not built.

[AjinkyaDahale]

Adding back the "do not merge" tag, because after rebasing with PR #5176 merged, the "front" face is not built.

This bug should now be fixed.

[freecadci]

for feature branch PR_5170. Pipeline 413453485 was triggered at f8da88c. All CI branches and pipelines.

[donovaly]

Excellent work.
I stress-tested the feature and it works nicely. I have only minor comments/clarifications.

[donovaly]

Many thanks for this PR. This improves the loft feature a lot since pyramids are not that rare (own experience).

Please have a look at my very minor comments. I will merge it then and then you can continue to bring Sweep/Pipe to the same functionality ;-)

(By the way, after sweep/pipe can handle vertices as well, I propose to make a PR to uniform the naming. At the moment the tern "sweep" is mixed with "pipe" in the code. We should only use one term and "pipe" is for non-native speakers like me even misleading.)

[donovaly]

I now found a test case that fails:

Here is the test file:
loftbug.zip

So the loft is nicely created but then not visible. Maybe this is because a body cannot contain of several volume shapes and in my test case they form one but the most thin point is a vertex?
However, this would not be a stopper for the PR.

[AjinkyaDahale]

I now found a test case that fails:
...
So the loft is nicely created but then not visible. Maybe this is because a body cannot contain of several volume shapes and in my test case they form one but the most thin point is a vertex? However, this would not be a stopper for the PR.

That is probably exactly what's happening. Could you try deleting the box and using a sketch instead?

[donovaly]

That is probably exactly what's happening. Could you try deleting the box and using a sketch instead?

This works, here an example where the begin and end is a single-point sketch:

However, what I wanted to test is if a vertex from a surface could be taken. Therefore i need an object like the box.

[AjinkyaDahale]

However, what I wanted to test is if a vertex from a surface could be taken. Therefore i need an object like the box.

Try one of the vertices on the left face. There should be some 2D (at least) intersection with the solid, or so I would imagine.

[donovaly]

Merging now, since the comments and wording changes are in.

The issue I mentioned in #5170 (comment)
is due to the fact that PD cannot have two 3D objects within one body yet. This is something @realthunder has in his branch, so in future this would be possible. For now, it is sufficient to mention the special case in the Wiki.