Although this is not the solution you wanted or expected, it is still technically a valid soliution. This issue shows up using the DogLeg and the BFGS solvers but not the LevenbergMarquardt solver.
Abdullah might be interested in this file for solver testing. I have sent him a message to find out.
Hypercube3D wrote: ↑
Sun Aug 23, 2020 12:59 pm
The constraints can be redone to avoid this as a work around.
e.g. instead of making the outer border 1mm from inner border, it could be made 6mm from center for example.
Yes, there are several other ways you could constrain this and in this specific example you could have created this sketch more efficiently as well. For example since in this case both squares are centred on origin, you could have just applied the symmetry constraint between diagonally opposed corners and origin, that way this would also work with rectangles and you would have only needed on symmetry constraint on the outer square. You also don't need any of the construction lines if you simply used another symmetry constraint on diagonally opposed corners of the smaller square as well. Distance constraint can be applied between two points on the end of lines without a construction line.
Also to make a more robust sketch you could just use two horizontal distance constraints, one on an edge of each square, then you have defined the inner square so add the same expression that you did to read its value from the spreadsheet, but for the outer square just use an expression that reads the inner square value then adds twice the border distance value, the Expression would be "Spreadsheet.border_distance * 2 + Spreadsheet.square_width"
OS: Ubuntu 18.04.5 LTS (KDE/plasma)
Word size of OS: 64-bit
Word size of FreeCAD: 64-bit
Build type: Release
Python version: 3.6.9
Qt version: 5.9.5
Coin version: 4.0.0a
OCC version: 7.3.0
Locale: English/Australia (en_AU)