=======================================================================
In uncommon situations, the analysis of QR decomposition leads to full rank, but the result does not converge.
We avoid marking a sketch as fully constrained when no convergence is achieved.
======================================================================
1. Instead of excluding internal alignment constraints from popularity candidate selection, exclude them from the group altogether. This
ensures no group is non-empty with uneligible candidates, which prevents an infinite loop in uncommon circumnstances (when DoFs collapse).
2. Ensure no internal alignment constraint is identified when conflict ensues.
The latter is intended for a specific solver constraint.
[Sketcher] Make further changes for tangent-at-knot separation
1. Remove code for tangent-at-knot from `addAngleAtPointConstraint`.
2. Use correct order of input.
3. Separate internal C0 knot vs end knots. The latter can still be constrained
but the user must use endpoints of the spline instead of knots.
Similar to 2715a66ff02a46f94ae3fc6527fd446e666b8e58.
Added some comments and removed some TODO's.
Return grad values directly rather than doing summation.
Also squashes:
[Sketcher] Make tangent-at-knot with just knot and line
[Sketcher] Disallow tangent at C0 knot
If passed on to planegcs can cause segmentation fault.
[Sketcher] (Re-)Support tangent at B-spline end-knots
New code had introduced problems for non-periodic spline end-points, and
periodic spline "end-points" were not supported anyway.
(here end-points mean star/end points)
It is possible to ask for splitting at an end point of the curve. This leads to
a `CADKernelError` and leaves us with a "hanging" clone. This check prevents that.
===================================
Changes from naked pointers to smart pointers are motivated to the use of functions that can reasonably throw under certain circumnstances (such as trim).
When introducing smart pointers, it is not necessary to explicitly delete the new geometry array at the end of the function.
When using the new facility to add a smart pointer geometry (previous commit), the copies generated in the split algorithm can be reused, which renders
keeping track of the new geometry for memory management unnecessary.
As geometry is added to the property which each call to addGeometry, the stored newIds can be reused if access is necessary to geometry pointers afterwards
(e.g. for constraint management).
================================================================
This new facility avoids to have to create a new copy() when a user copy is already created.
As the user copy is reused via move semantics, memory management is simplified.
CAVEAT: When this facility is used, the client code has to ensure whether a copy() or a clone() of the Part::Geometry
should be undertaken. The different between both is that the former creates a new uuid (tag), whereas the latter does not.
=========================================================================
A migrated parabola cannot be openned with a previous version of FreeCAD. The user is notified upfront.
Algorithm to join b-splines:
The code simple concatenates the knots, poles, weights, and knot multiplicities
together, removing data on the connection point of the second curve. Some
further study is needed to see if/when it will give an exact/good connection.
Icon courtesy @bitacovir.
Comments by @abdullahtahiriyo.
Remove default values and smaller constructor for `CenterOfGravity` and
`WeightedLinearCombination` constraints.
Clarify comments.
Improve readability of `CenterOfGravity` and `WeightedLinearCombination`
constraints.
Also squashes:
[Sketcher] Create center of gravity constraint in planegcs
[Sketcher] typo
[Sketcher] Use accurate "weights" for knots
By weights we mean the linear combination factor B_i(x) such that
spline(x) = sum(pole_i * B_i(x)) for _non-rational_ splines.
[Sketcher] Use more appropriate weights for knots
These are relevant for knots _away_ from any ends (and possibly other high
multiplicity knots).
[Sketcher] Make COG constraint weights user-definable
[Sketcher] Make `flattenedknots` for periodic B-Splines
[Sketcher] Fix incorrect setup of `flattenedknots`
Without ensuring enough space, iterators become invalid. These iterators are
needed because for periodic B-splines we need to pad flattenedknots with offset
values within flattenedknots.
Apparently there is still some iterator issues even after the reserve. Just use
fresh vectors instead.
[Sketcher] Apply knot constraints by parameter
Hopefully this will allow directly applying constraints on knots.
[Sketcher] Disable some knot updating
[Sketcher] Use center of gravity constraint on knots
[Sketcher] Fix knot COG constraint for periodic splines
[Sketcher] Add start/end point of periodic spline to solver
This removes the trouble of transferring constraints to the underlying knot.
[Sketcher] Support knot constraints on rational B-splines
[Sketcher] Remove virtual from overridden methods in planegcs
Follow 0penbrain's comments
[Sketcher][planegcs] Use `unsigned int` in signatures
Also `size_t` at places
Suggestions by @abdullahtahiriyo
================================================================
Master has a problem in that internal alignment constraints are suggested to the user for removal.
This is fundamentally wrong, as an internal alignment constraint are an inherent part of the geometry. They cannot be the ones suggested for removal.
The popularity contest algorithm is an heuristic algorithm that determines which redundant/conflicting constraints should be proposed for removal.
Basically, the algorithm works on groups of redundant/conflicting constraints detected via the QR decomposition. A constraint may belong to more than one group.
The algorithm runs some heuristics, each constraint scoring a value, the one constraint from each group scoring the highest is proposed (is more popular and wins the contest).
This PR documents the algorithm, and adds a further condition, that internal alignment constraints are never proposed.
As the solver works with solver constraints as opposed to the sketcher, which works with sketcher constraints, information about whether a solver constraint originated from a
sketcher constraint that is internal alignment is necessary. So the solver constraint is extended to accomodate this piece of information.
As a bonus, it fixes a bug. Solver constraints carry information of the ID of the corresponding sketcher constraint in their tag. Knots are not currently implemented as constraints.
However, the tag index was not being update. This caused the popularity contest to provide wrong suggestions despite good detection.
