and in the case of a singular Jacobian, report which constraints
can be removed to fix it. Also a mechanism to hover and select
entities and constraints from the text window.
[git-p4: depot-paths = "//depot/solvespace/": change = 1746]
triangle list before adding to the BSP, and check if a point in
plane lies in multiple triangles and choose the one with the
biggest normal magnitude (i.e., area) when testing normal
direction.
[git-p4: depot-paths = "//depot/solvespace/": change = 1745]
implement that. Also make solver work only between the first and
last visible group; earlier can just work from previous solve
result, and later don't matter.
There's some issues with the csg code; it will eventually produce
an open mesh, which is very bad. Not sure whether that's a logic
bug, or a numerical issue; still generating absurd triangles pretty
routinely.
[git-p4: depot-paths = "//depot/solvespace/": change = 1741]
polygon, not just triangles. This helps to avoid needless
splitting. Also test if an entire triangle got inserted in multiple
pieces; if yes, back things out, and just insert the triangle.
Also remove the extra partition stuff, since it didn't seem to help
consistently, and this does.
Still could do some better merging, in the case where an inserted
triangle does not get fully inserted, but we can find a better
triangulation than what the BSP naturally gives.
[git-p4: depot-paths = "//depot/solvespace/": change = 1739]
zero-area triangles, apparently. And make the number of line
segments used to approximate a triangle depend on its scale on
screen.
[git-p4: depot-paths = "//depot/solvespace/": change = 1738]
while building it. That may improve performance, by building a more
balanced tree and actually reducing splitting. Not dramatic
improvements, though; half the triangles for some parts, but no
change or slightly worse for others.
[git-p4: depot-paths = "//depot/solvespace/": change = 1737]
set of coplanar faces. The polygon count still gets stupid fast;
I'm thinking I can fix that by adding some extra test planes at the
top of the 3d BSP, to quickly cull out stuff that doesn't intersect
us.
[git-p4: depot-paths = "//depot/solvespace/": change = 1736]
a triangle mesh in a BSP. That works, although it splits too often,
the initial triangulations are not good quality, and coplanar faces
are not yet handled. I'll do the coplanar thing tomorrow.
[git-p4: depot-paths = "//depot/solvespace/": change = 1735]
for that, and storing the triangles instead of rendering them
immediately. Not sure if that's smart; in theory could change from
implementation to implementation, but the results look much better
than I would get myself.
[git-p4: depot-paths = "//depot/solvespace/": change = 1733]
not very well; I'm doing a b-rep, where the boundaries are complex
polygons, and there's too many special cases. I should probably
replace this with a triangle mesh solution.
[git-p4: depot-paths = "//depot/solvespace/": change = 1731]
that's in a group that solves after the current group. This means
that the constraint/request gets deleted instantly, and stuff like
auto-constraints on the just-created entities blow up. I now check
that the active workplane comes from the active group or one solved
before it.
[git-p4: depot-paths = "//depot/solvespace/": change = 1729]
as a constraint on the direction cosine, rather than driving the
dot product against a rotated vector to zero. The drawing is the
ugly part; to do that for skew lines, I gave up.
Also add a function to clear non-existent items on the selection
after solving, since that could have caused an oops().
[git-p4: depot-paths = "//depot/solvespace/": change = 1727]
the coordinate system (x, y, z normal vectors) in the bottom left
corner of the screen at all times, and hide group-created
workplanes except when that group is active, and activate that
workplane when the group is activated.
[git-p4: depot-paths = "//depot/solvespace/": change = 1726]
sketch (e.g., a line whose length is constrained gets deleted, but
the constraint is left behind; or the point that's the origin for a
drawing group in plane gets deleted), then deleted the dependencies
too.
[git-p4: depot-paths = "//depot/solvespace/": change = 1725]
everything broke; apparently that driver didn't like me continually
destroying and recreating the HPGL context, and it also didn't like
me drawing zero-area polygons for my edges (which seemed like a
good idea, because it let me use glPolygonOffset instead of doing
that by hand). So it now all seems to work again, and faster.
[git-p4: depot-paths = "//depot/solvespace/": change = 1723]
use the horizontal or vertical axes of the active workplane. The
equations that we write from these are very fast (of the form u + v
= 0, or u - v = 0).
[git-p4: depot-paths = "//depot/solvespace/": change = 1720]
lets me reweight the parameters, e.g. to encourage the solver to
make changes to the point being dragged.
[git-p4: depot-paths = "//depot/solvespace/": change = 1719]
we need something to force the points into plane, and the workplane
supplies that), but otherwise straightforward. And add diameter and
equal radius constraints for the arc.
[git-p4: depot-paths = "//depot/solvespace/": change = 1718]
solve the Newton's method iterations in a least squares sense. This
is much less likely to result in numerical disaster; a parameter
will never make a very large change, unless that really is the only
way to satisfy the constraints.
[git-p4: depot-paths = "//depot/solvespace/": change = 1717]
list, and then adding a new entity to that list, and then looking
at that pointer again. Not okay; the add operation might have
forced a realloc. I have to watch for that.
And add a "distance ratio" constraint, plus a new kind of group
that comes with its own workplane. The workplane is not solved for;
it's generated explicitly in terms of elements that are already
solved.
[git-p4: depot-paths = "//depot/solvespace/": change = 1716]
translation; or equivalently, rotation about an arbitrary axis).
Those will be important for step and repeats, and for imported
parts.
Also fix a terrible memory corruption bug: I was freeing the remap
list after I loaded it from the file, but the code that put that
into the SS.group list made only a shallow copy.
[git-p4: depot-paths = "//depot/solvespace/": change = 1715]
constraints work mod 180 degrees, so that it snaps to however the
workplane was drawn (more vertical vs. more horizontal).
[git-p4: depot-paths = "//depot/solvespace/": change = 1714]
constraints. And generate the constraint equations for entities
(e.g., that our unit quaternions have magnitude one). Numerical
troubles there, but it sort of works. Also some stuff to draw
projection lines with projected constraints, and to auto-insert
more constraints as you draw.
[git-p4: depot-paths = "//depot/solvespace/": change = 1711]
remap when I copy circle entities, in order to make the radius
numerical somehow (analogy with the POINT_ and NORMAL_XFRMD) thing.
[git-p4: depot-paths = "//depot/solvespace/": change = 1710]
segments), add the toggle construction command, and color the lines
differently depending on what group you're in.
Also change dynamic memory stuff to use a Win32 heap for everything
(no malloc), and validate that often. I think I've seen it crash,
though I can't reproduce it.
[git-p4: depot-paths = "//depot/solvespace/": change = 1708]
hadn't previously noticed, because I didn't use to have workplanes
with non-zero offsets. And clean up the interface to normals a bit.
[git-p4: depot-paths = "//depot/solvespace/": change = 1707]
workplanes. And fix up our polygon normals, so that everything gets
shaded correctly (and so that later we can generate our STL files
with correct normals).
[git-p4: depot-paths = "//depot/solvespace/": change = 1706]
vectors", represented by unit quaternions. This permits me to add
circles, where the normal defines the plane of the circle.
Still many things painful. The interface for editing normals is not
so intuitive, and it's not yet clear how I would e.g. export a
circle entity and recreate it properly, since that entity has a
param not associated with a normal or point.
And the transformed points/normals do not yet support rotations.
That will be necessary soon.
[git-p4: depot-paths = "//depot/solvespace/": change = 1705]
diagonal elements of the matrix summed to -1. Now it's ugly, but I
think that it's correct. And add a command to flip the view to the
other side, which is what started my problems. And tweak display of
H and V and M for constraints: put them in the constraint plane, so
that they're stationary as you rotate around.
[git-p4: depot-paths = "//depot/solvespace/": change = 1704]
faces of the polyhedron. And shade the faces when I draw them, and
fix up our projection matrix so that the depth testing works
properly.
[git-p4: depot-paths = "//depot/solvespace/": change = 1703]
the point that's being dragged first, to guarantee that that one
gets the max possible degrees of freedom. The sort code (sort a
list of integers, then apply the permutations by swaps) was more
painful than it should have been.
[git-p4: depot-paths = "//depot/solvespace/": change = 1700]
turned out straightforward, in great part because the planes are
workplanes (6 DOF, represented by a unit quaternion and a point),
and therefore make it easy to get a vector in the plane, as well as
a normal.
And on that subject, replace the previous hack for parallel vector
constraints with a better hack: pivot on the initial numerical
guess, to choose which components of the cross product to drive to
zero. Ugly, but I think that will be as robust as I can get.
[git-p4: depot-paths = "//depot/solvespace/": change = 1699]
or taking partials (constant folding). Also keep a little hash
table to mark with params are used in each equation, in order to
quickly discard trivial partial derivatives. This is solving a
64x64 system in <20 ms. I suspect this is now much faster than
Sketchflat.
Slightly fake situation, though, since substitution solver has not
yet been written, and no partitioning. I'll do those next.
[git-p4: depot-paths = "//depot/solvespace/": change = 1698]
not have much motivation behind them, but they seem to work. And
make sure that we don't solve multiple times without repainting in
between, and tweak the text window a bit more.
[git-p4: depot-paths = "//depot/solvespace/": change = 1696]
Jonathan Westhues