[OpenSCAD] Fusion360 and OpenSCAD
doug at moens.org
Tue Oct 4 18:49:55 EDT 2016
Marius said: "It would be interesting to look into isomorphic
representations, where text-based and a flow graph could just be two
different editing modes."
I've thought about this too, although from my perspective it would be
difficult to pull off[*]. Still, if someone manages to do it, it could be
[*] The difficulty is that the text based representation might contain
information that isn't contained in the graphical flow chart, and vice
versa, so what does the file format look like, how human readable is it,
what happens to this information when you switch between views in the
editor. An Antimony-like flow graph might contain X-Y coordinates for each
block. If the flow graph is a directed acyclic graph, as in Richard's
writeup, then the text version will need to introduce variable names and
`let` blocks that aren't needed by the graphical view.
Antimony avoids these issues because the flow graph and the programming
language representation are two different file formats.
As for SISAL, my impression is that it doesn't have too much to do with
Antimony-style visual programming. Instead, I view it as "functional
programming for Fortran programmers". In SISAL, there is a big emphasis on
compiler technology (automatic parallelization for SMP architecture
machines and efficient code generation), and there is also a lot of
language design focused on providing a convenient and efficient way to
write loops for array based code without using recursion or other
functional programming design patterns. But SISAL is still a functional
language (in the same sense that OpenSCAD is): functions are pure and
referentially transparent, with no side effects; there are no mutable
global variables; variables are single-assignment.
On 4 October 2016 at 16:29, Marius Kintel <marius at kintel.net> wrote:
> > On Oct 3, 2016, at 11:19, Richard Urwin <soronlin+openscad at googlemail.
> com> wrote:
> > Have you considered that OpenSCAD is to all external purposes a Data Flow
> > language that bears striking similarities to languages such as SISAL? (In
> > fact making OpenSCAD more like SISAL would be wonderful -- there's more
> > orthogonality and more powerful structures there, but little chance of it
> > happening.)
> I’ve been looking into ways of using OpenSCAD as a dataflow language, but
> I haven’t looked at SISAL in particular.
> Most of the concepts in OpenSCAD should be pretty well mappable to a
> dataflow or flow-based programming graph. It would be interesting to look
> into isomorphic representations, where text-based and a flow graph could
> just be two different editing modes.
> So far I’ve temporarily shelved all of this in search of more resources to
> get this off the ground, but take a look at e.g. Antimony, which has
> combined flow-based programming with an f-rep kernel to achieve smth. very
> OpenSCAD mailing list
> Discuss at lists.openscad.org
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Discuss