Fast FreeCAD progress

[andremiv]

Hello Dear Freecad community,

As you know, I am developping a C CAD, therefore completly different than FreeCAD that I call Fast FreeCAD.

Libs :

I will use modern opengl with glfw for context creation. I will write to image and use it in the obsolete gtk+ with cairo (gtk2) library for gui. Cairo will be used for text blitting outside of opengl. Clearly not the best integration of the opengl context but I feared to use gtk3 and gtk4 due to bad press. Qt is c++ based so it is out of question for me. Should I use gtk3 or gtk4 instead ? I read that it is very slow, the obsolete library gtk+ is fast. What do you think ?

Then GNU MP for the non linear solvers, and probably various solvers that I will copy. The aim is to support implicit and parametric surfaces, curves and their intersections. I want to reduce the size the FFCAD project take, I think FreeCAD projects takes too much disk space, each project should create a subdir where useful intermediates are stored but not within the project since it can be generated. I want CAD with decent algebraic support, FreeCAD suck in this domain, slowing down the whole process.

Vim will be used as the editor of the script as openscad. I will try to autogenerate the gui.

Then you also have open mpi for parralel computing, and ATLAS BLAS + extra sparse solvers like SUPERLU.

You also have the object pointer/selection ray tracer, and the various shaders in glsl for displaying and dividing the curve

The objective of the Fast FreeCAD is also to automate the meshing and distributing of the meshing for fast resolutions of the PDEs.

In essence Fast FreeCAD project is to make a software like Freefem++ but with better 3d CAD capabilities and better automations. Yes the word better is the best word to describe the project.

I will say that I am just about finishing grasping the difficulty of CAD, meshers and pde solvers in general. They are all linked in some way, and no free software is good enough for scientific or engineering purposes, though FreeCAD was a good step in the right direction. I am more than 100% sure the occt is bad kernel and they do it purportedly so you buy their better kernel

See you soon (less than a year), I will keep you updated. Of course this will be Free software so you will be able to demo it for the curious with a github clone.


Miville André

[gabokamaze]

completly different than FreeCAD

but

call Fast FreeCAD

... I do not get this ... And can you call something FreeCAD without any consent from anybody ? I remember that I saw few months ago a topics dealing with using FreeCAD name and logo which, in a way, is now "protected" or something like that -> am I wrong ? (edit : I regained it : https://forum.freecadweb.org/viewtopic.php?f=9&t=72335 )

Should I use gtk3 or gtk4 instead ?

If you want something fast, take a look on FLTK : https://www.fltk.org/

Vim will be used as the editor of the script

Is it a personal project or is it intended to be used by "designers" ? For the latter, Vim is clearly not a good choice (and I am Vim user )

I am more than 100% sure the occt is bad kernel

I am not an expert, but it looks like occt is the open source kernel the most widely used, this is not probably a hasard.

Regards

[onekk]

There is a lack of geometric kernel around, BRL-CAD is using a different kernel based on NURBS, but as you speak about mesh and triangulations this is not resemble a CAD as real objects are not triangulated.

Using old, deprecated and unmaintained libs like gtk prior to 3 I think this is not a clever move with future in mind.

But as C++ is according to many statistic the most used language even avoiding C++ as devil will not attract many developers.


About CAD kernels, in the past there have been many failed attempt to make an opensource CAD and there are not OpenSource CAD kernels out of here it could be seen as a demonstration that not is very easy to develop a CAD kernel.

So good luck and good work.

Side note:

Probably you will incurr in some problem using as part of the name of your future CAD FreeCAD as probably is copyrighted by the FreeCAD Project Association (FPA) (sorry if I have remembered wrong the name)

@sliptonic @yorik wath do you say about the use of FreeCAD?

Regards

Carlo D.

[johnwang]

Maybe FFCad which is free fast cad.

[chrisb]

I don't think that the name is here the biggest stone to be rolled up the hill. So let's see how the rest progresses. There sure are cases where a speedup would be highly appreciated such as Patterns and Arrays. I'm pretty sure that speeding up these would need a new kernel.

[obelisk79]

I feel like the naming is a bit awkward, as well as it seems odd to me that one would post about an entirely new CAD in FreeCAD's development forum.

All that aside, I'd like to comment about the OCCT being the only opensource geometric kernel. I'm sure it's not feature complete yet, however there is a relative newcomer under active development in this field called "Truck" written in rust. https://github.com/ricosjp/truck
This seems to be funded/driven largely by Japan's "Research Institute for Computational Science"

It seems to have real potential to supplant OCCT as the best open source kernel available. It would be nice to have an alternative that can fillet properly.

[chrisb]

It would be nice to have an alternative that can fillet properly.

Thickness working on B-spline curves (sigh), tapered B-spline extrudes (SIGH).