this is not as easy as it sounds as there are several aspects going on.

1) We need to have a small group of developer that can work in sprint’s to design several aspect’s of any new codebase so we have something running in a small amouth of time, this can set the base for further developments.

2) Somebody needs to write a design document ao any developer within the team can follow this design.

3) Somebody needs to write a architectural document so everybody can understand how to develop against LibreCAD, so we don’t get a hack’ish design and put’s us in the same position as we are currently at.

before we do anything, 2) and 3) are very important. I personally have a idea how thing’s should be designed, Dli and I am sure some hard-core c++ user with CAD experience have their own ideas!

The problem is to get everybody on the same line and make time to have this happen...

2 years ago I did develop this : https://github.com/rvt/kerneldev

It was designed with the following in mind:

1) Separate of concern.

 * Application (the GUI) has it’s own codebase

 * The kernel (that handles CAD entities) has it’s own codebase

 * a CAD viewer /that handles viewing of entities) has it’w own codebase.

2) All CAD entities are immutable it has the following advantages

 * Very easy to handle unto/redo

 * Multi processor ‘ready’

 * Easer to make a storage layer (DB, File, REST etc…)

3) ‘Kernel’ handles all entities and it’s modifications through a clear interface

 * This will make plugging various things into the kernel a LOT easer without having the modify the viewer or the GUI

Here is a bit of code example to create lines:

The only way to get your drawing modified is using the operateOn method. So even if some code tries to ‘move’ a line/circle etc,

the kernel will never be updated you cannot do this on line , example : myLine=line->Move(10,0); 

…

...