There is a valid point to say this is not the right way to do it for softbodies. Cloth simulation already has air pressure, why add air pressure to softbodies as well? Softbodies are solid chunks of elastic or deformable material, not air tanks as it might seems currently. Blender really has very bad softbody code so using air to fix the volume problem doesn't really cut it. In my opinion blenders softbodies are made redundant by cloth simulations currently and would need a proper re-implementation with a finite element solver. The only thing softbodies have, that cloth doesn't have currently is plasticity, but that may very well come to cloths simulations in the future (for aluminum foil simulations for example)

What about the new mantaflow simulation, cloth simulation and softbody simulation? All of those are missing units. (I get that they don't really do physically correct stuff but there could be an effort to make it understandable to a physicist looking at it)

• fix negative thickness

Added one more fix for angle clamp in complex mode to this revision. I don't know though if it's so smart to put everything in one revision... (anyway)

quick refractor to make the code a bit more beautiful and awesome before commit (clang, order of operations, 1 less if-condition)

this button should extend the selection to the armature of the (still active) object and present a mode similar to face/verts select in that there is a selection tool in the toolbar to select bones (you can do it with shift LMB but the problem here is the shortcut ambiguity, so you can not select a single bone but only add/remove a bone from selection, just like with face/verts selection)

This is now implemented in Blender with the new Weld Modifier.

I have written temporary python code to try that, when I made the new Solidify Complex mode. I plan on implementing intersection handling into solidify eventually like I did it in my python script test code. Try to write everything in python with bmesh at first, before you dive into modifiers, it makes work a lot easier to know what works first. You can take a look at my naive approach at https://github.com/redweasel/manifold-solidify

now it can be reviewed

I want to make offset = 1 and offset = -1 also more consitent before this lands, but there its a bit more involved than changing a number

• changed the names of the modes for the user (left the internal names the same to avoid collision of the word SIMPLE for method and mode)
• applied clang formatting

Here is an example file with all the objects from development and fixing.
The name Manifold comes from it being able to make non Manifold geometry into completly manifold shells.
I don't know how "Intersect" would fit as a description, because that is exactly what it is missing.

• fixed previously unhandled open singularities
• fixed miscounting of unused loops for fill faces
• fixed problems with singularities at rims
• Fixed miscalculation of loops for closed splits
• removed printf's

Should be ready for review now again.
EDIT: (forgot to apply clang formatting, will fix that with the next revision or before it lands)

I still found a bug, but your file is fixed now. The new one comes in when you extrude you mesh without translating afterwards, creating duplicate geometry.

• fixed previously unhandled open singularities
• fixed miscounting of unused loops for fill faces

• fixed segmentation fault

ok I will stop saying "fixed all bugs" just fixed some bugs

• finally fixed everything (a lot and I forgot about most things already)

fixes include:

• handling of singularities (as I call verts which have a one-sided faceloop around them)
• fix duplicate edge and poly handling (still not recommended to use but the modifier wont fail you)
• various other smaller bugs

• clamped mat_nr (could go out of range before)
• fixed clamp angle segmentation fault

I think the revision is finished now. I didn't find any more bugs.

• check for offset == 0 for extrude mode
• there is a segmentation fault that occurs sometimes when using extrude, even thickness, angle clamp. It seems to appear at no specific point, just after using it a while (like for animation)

• cleaned up code a little
• fixed isolated verts in result for only rim

• removed clamps for cosf in confusion why I ever added them (cosf is defined for all finite values of float)

• fixed assertion error (had an outdated condition)
• fixed size of vert_adj_edges (was numEdges instead of numVerts) and fix for isolated verts
• added clamping for cosf everywhere where I used it

fixed all sort of linear algebra mistakes and cleaned up the angle calculation.

improved naming further, used a few more makros, fixed constraints calculation for offset != 0. Need to fix even and simple method for offset != 0 as well but had no time for that yet. The error in solidify_crash.blend is still unresolved.

made naming more understandable, applied all the other suggestions from above.

Ok I have fixed all the bugs I found and added a lot of code to support duplicates. Also I added the offset property that was missing. I have a file with my test objects that I will add to this comment.

fixed a bug where very complicated geometry would create wrong loops resulting in all sorts of failure

fixed normalized vertex normals for curves

changed the obsolete designated initializers to standard syntax

As I understand the apply operators, they should all leave the visuals exactly as is, so this is a bug. It completly not doing anything with axis angle representation is a bug for sure though, but even the problem with the quaternion isn't, that it rotates the cube, the problem is, that the resulting delta quaternion is not a unit quaternion (which all rotation quaternions are unit quaternions)