Might be my limited English too...sorry about that.
Ok. Let's start from scratch...
You might be aware of the difference of Pixel- and Vectorgrahics in 2D?:
- Pixelgraphics (Paint apps) create an image by colored squares (Pixels) as you may know it from photography. If you zoom in such an image you'll soon see these squares getting bigger and bigger.
- Vectorgraphics (Draw apps) instead draw an image by using math functions. A circle is drawn based on a centre, a radius and a certain line thickness. If you zoom into such an image representation you'll never find some pixel stairs. The line stays smooth and keeps the same thickness.
These two ways of graphics exist in 3D as well:
- There's a representation describing a 3D shape via a net of planar polygons (maybe the simplest polygons: triangles) analog to pixel graphics. Depending on the mesh density you'll find those elements zooming in sooner or later. A sphere is made of a certain number of flat polygons. This representation's quality depends on the number of used polygons and will never (for it is made from planar elements) be a perfect sphere.
- There's a way of representation describing a 3D shape via 3-dimensional math functions. This way is used by construction applications ("CAD" - basically mesh processing is CAD as well, but the math attempt occupies this term) where the construction parameters are known. Here the shapes are perfect - a sphere can be done just by a centre and a radius.
>>> First basic insights:
- Both ways need different ways to create or modify something.
- Both own different powers.
- While you easily can convert a math based body to a mesh representation (by dropping it onto a certain grid), its hard to find math rules from a mesh.
- You can't directly transfer needed operations from one representation to the other.
>>>>> Note at machining: Generating a toolpath from a mesh ends in (more or less) linear travels. A waterline path around a mesh-sphere gives as much travels as triangles hit by the current slicing plane. Each needs acceleration and deceleration. So the travels might be that short that the machine will never reach its highest speed. Processing a math-based representation instead can do a path around a sphere in only four non linear travels. So: Expect a longer machining time on meshes.
Now there are things which are hard to describe via maths and there's no choice which way to go: Maybe the source is already a mesh (as got from a 3D scan), maybe the object is too complex to describe it via maths: You simply need mesh processing.
MM deals with meshes only. Let's look at the way of meshes more closely:
A mesh is just a surface ( a Shell) without any stuff within - like an air balloon. The balloon's outside is of a nice white, while its inside owns an ugly pink-Barbie-like stripe pattern. You hate Barbies and you don't want to see that pattern. So you fill the balloon with sand (just in case it bursts). You've got a watertight volume filled.
Now that balloon+sand is too heavy: You decide to put another balloon (Shell) into the first. But wait, you hate the pink stripes and you might see it through the air outlet: Better to flip it so it's inside out. Now the stripes point to the sand and are hidden. Congrats!
>>> Second basic insight: On a mesh there are front faces and back faces. Back faces point towards a (virtually not existing) material. If you want a watertight cavity within a shell you'll need to define a second watertight shell and flip it inside out (to flip an object: SelectAll of it and do Edit/FlipNormals; to combine two objects: Activate a second object by clicking on it while holding down Shift and do EDIT/Combine) .
>>>>> Note on a cavities in CAD: This flipping doesn't work. You need to actually do a boolean subtraction over there.
If you're using a flipped cylinder and Combine it, you can create a cavity within an object which will be visible as a hole if you cut the object later on.
To create a fitting dowel rod you simply need a copy of the cylinder. Create this copy (via the double-square button in the ObjectBrowser) before flipping and combining it. You may hide this "male" part (via eye icon in ObjectBrowser) for now.
Ok. So far for today. It's late at night over here right now... About a slicing workflow tomorrow..