OK, I've re-read that and looked at the workflow using the "Part" and "Part Design" workbenches. I'm a bit disturbed that the result of the "Part Design" method is similarly linear to what I'd ended up with, and that the description shows that as effectively being the only way to work (with that workbench).MarkMLl wrote: ↑Thu Jan 27, 2022 12:00 pmThanks, I'll follow up.drmacro wrote: ↑Sun Jan 23, 2022 11:21 am Have you read and worked the examples on this page?: https://wiki.freecadweb.org/Part_and_PartDesign
The examples down the page create identical models exclusively in Part workbench and Part Design workbench.
Apart from that, I've been experimenting with "Polar Pattern" etc. and would appreciate confirmation of a couple of things if anybody could spare the time.
First, when making a pad one starts off with a single sketch, there is no effective way of merging two sketches.
The sketch has to have an unbroken outline and no connected internal lines, e.g. has to be a single rectangle rather than two rectangles even if two pairs of points were constrained to be coincident. If there are connected internal lines, the result is a TopoDS::Face error.
It's not possible to connect two pads/pockets and then make a polar pattern from the result. For example, a strengthening web (as shown in the wiki example) with a hole bored through to save weight and material.
However if the sketch has unconnected internal detail, e.g. comprises a rectangle with a circle drawn internally, then the usual topological conventions apply and when padded the circle becomes a hole and the result may be polar-patterned.
Parts produced by the Part Design workbench and solids produced by the Part workbench may be fused by Part workbench boolean operations, with the top-level result being e.g. convertable to a mesh hence a .stl.
However so far it appears- to me- that parts have to be specified and in particular positioned numerically. I'm troubled about situations where the maths doesn't work out nicely, which is why my original example had a seven-sided prism.
If an entity has to be e.g. positioned on a specified face of another entity, this has to be done at this level.
With my current level of understanding, I'd hate to attempt e.g. the balustrades on a spiral staircase. (I admit, as soon as I wrote that I realised that drawing the staircase and then positioning the balustrade on each tread wasn't the way to do it. But I think that that sort of positioning job is still a reasonable illustration.)
I'm still very uncertain about the distinction between parts and bodies etc., and the significance of e.g. a fusion operation appearing outside the current part.