Here is a long thinking path, so try reading the Proposed solution at the end...
The main use of them is to save the time:
-- decompose a huge solver's equation onto smaller
-- avoid preparing jobs like reference DatumPlanes
-- avoid binding jobs by importing external geometry
But also the problems that might come up:
-- if there is an "import" link from one layer to another and the original of imported element is deleted: what happens with the dependant line?
So on implementing layering, might imagine:
-- solver toggle control
-- a specific naming mechanism
-- line replace tool (to auto-re-aplly constraints to a new element) etc.
Might be an easier solution (from the view to simplify solving) to implement groups in the sketcher
And a control to perform solving on all groups or just on a selected group.
( groups can be hidden, shown, toggled sketch/construction state,
imported as ShapeBinders to other bodies
(??changed inner/outer display colour), )
There would be a TotalSketchGroup, and a tree of sub groups.
Groups might be fixed - thus excluded from recalculation.
The same functionality is applecable to Layers. Layers would serve as grouping container.
No building a fence. With multiple vertical rods. Each rod must be positioned. and Guide line requires 3 constraints + 6 contraints for guide
Having 50 rods on one section it comes to 450 constraints. It slows down the modelling.
Just before reading this topic - have come to the solution to split in "pseudo-layers" - multiple sketches.
It gave me two sketches(SkB and SkC) (instead of one possible) dependant on the third (SkA)
So on each change of the SkA I have to correct two sketches (open-edit-close)*2.
The opposite, what is also mentioned here - if there happens destruction of the sketch SkC - due to changes in the SkB - possible to delete SkC and make it again. While if it was one single sketch - it would break in whole (?). And would be a need to re-make all the "layers".
So maybe the solution would be to implement a "outer layering" - attach sketch to sketch?
And make it possible to switch between sketches trough TreeView without closing one sketch and opening another?
Seems the only advantage of layering is: easy switching between editing of different features.
So... the grouping element for sketches would be a SketchGroupDatumPlane. And all sketches attached to it are editable like layers with easier switching between them?
After reading some more in this topic, it seems that groups (maybe even not layers) would be more efficient and useful to limit the area of equation for the solver.
And also looks nice to combine the "outer-layering" and internal grouping mechanisms - these would complement each to another.
And this would involve mainly the TreeView programming
\\ \\ ExternalGroup:= MainFrame
\\ \\ Group_InnerFrameOffsetConstructional
\\ \\ Group_InnerFrame
\\ \\ ExternalGroup:= ImportInnerFrame
\\ \\ Group_OrnamentFramesConstructional
\\ \\ Group_OrnamentFrames
\\ \\ Group_Ornament_1_Elements
\\ \\ Group_Ornament_2_Elements
\\ \\ Group_Ornament_N_Elements
After having the A2Plus freezed ( no conflicts found, but no solving happens on applied constraints)
And not possible to locate the problem - seems that in-file/in-object module-limited solving would be useful all-around the FC.