Sorry about the negative tone of my response. I did get frustrated by your
original post, though. It's written in a pseudo-code, somewhere between
Lisp and OpenSCAD. Since you weren't proposing that as the final syntax, I
tried to translate your examples into OpenSCAD syntax, and what I finally
ended up with were list comprehensions + Haskell-inspired list functions.
Which is not what you were proposing.

The entire Genera OS was written in lisp on a lisp machine. Yes an entire

I never had an opportunity to use a Lisp Machine. I'm a student of
computer history, though, and it's legendary. I'd like to spend time
playing with this software so that I can understand why it's the best
development system ever. Maybe somebody will resurrect this in an emulator
one day.

I completely agree with adding structured assignment. I've been writing
OpenSCAD2 code as part of a project, and I've added structured assignment,
because it is so convenient for destructuring 2-vectors and 3-vectors in
the style of geometric programming I'm playing with.

Your example is pseudo code, so I'll translate to OpenSCAD syntax:
[ for ([a,b] = [[1,2], [3,4], [5,6]]) [b,a] ]

In my version of structured assignment,
[a,b] = x
produces a run-time error if x is not a list of length 2. The syntax works
in definitions, function formal parameter lists, and for statements.

You intend this to mean 2 loops running in parallel, instead of being
nested. In OpenSCAD, the syntax
for (i=[1:10], j=[2:11]) ...
looks like it would be good syntax for two parallel loops, but it's not.
It actually means nested loops, same as
for (i=[1:10]) for (j=[2:11]) ...
So I think there is a backward compatibility problem with expressing
parallel loops in the way you suggest. The problem is in violating the
expectations of OpenSCAD programmers, who expect 'for (i=.., j=..)' to mean
nested loops.

Your previous post had several examples of parallel loops. Here's one:

• Create pairs
  [for x = [1:5]
  for y = ["A","B","C","D","E"]
  do [x y]  ]
  giving  [[1 A]  [2 B]  [3 C]  [4 D]  [5 E]]

This is more pseudo-code. The problem is that if we convert this into
OpenSCAD syntax, we get nested loops:

[for (x = [1:5])
for (y = ["A","B","C","D","E"])
[x,y]]
and the result is a list of 25 pairs.

In functional style programming, you would create the pairs by calling
transpose, which in OpenSCAD is matrix transpose from linear algebra.
transpose([[1:5], ["A","B","C","D","E"]])
=> [[1,"A"], [2,"B"], [3,"C"], [4,"D"], [5,"E"]]

(This works in OpenSCAD2 because [1:5] is operationally equivalent to
[1,2,3,4,5] in all contexts.)

We could invent a new syntax for parallel loops in OpenSCAD. Maybe this
would work:
for (x = xlist | y = ylist) ...

Or we could just use transpose. Because this also works:
for ([x,y] = transpose(xlist, ylist)) ...

• loop terminatoin with while, or until, or just one and use not operator

I'm not sure about the , before the while, for reasons previously
discussed. And the argument of 'while' should be parenthesized, for
consistency with the rest of OpenSCAD syntax. So maybe this:

 [ for (x=[1,2,3,4,5,6] | while(prime(x))) [x, x*x] ]

In functional programming, you'd implement this using take_while, which
returns the longest prefix of a sequence that matches a predicate. In
OpenSCAD2, the code would look like this:

 [ for (x=[1,2,3,4,5,6] >> take_while(prime)) [x, x*x] ]

It's not much worse, and no additional syntax is added to the language,
since take_while is a library function. There is also a drop_while function
which discards initial elements that match the predicate, and I'm not sure
if this is expressible using the LOOP macro.

• define more internal loop variables. inner scope only

But we already have syntax for this.
[for (x=[1,2,3,4,5,6]) let (start=12) x*start ]

I had the same problem, reading your previous post. You wrote:

• using conditionals within the loop (as an operator, not a regular if
  statement)
  [for x below 10
  if odd(x)
  collect x into odds
  else
  collect x into evens
  finally [odds, evens] ]
  gives [[1 3 5 7 9]  [0 2 4 6 8]]

In functional programming, this would be done using 'partition'.
partition(odd)([0:9])

Here is the non-recursive definition:
partition(f)(s) =
let (p1 = [for (x=s) if (f(x)) x],
p2 = [for (x=s) if (!f(x)) x])
[p1,p2];

In summary, my counter-proposal is to add structured assignment to
OpenSCAD2, and then provide library functions to make writing loops and
list processing more convenient.

I'll admit I have a bias for the functional style of programming, since
I'm used to that, and I don't have any experience using the Lisp LOOP
macros. And I admit there's a learning curve; you need a cookbook to learn
some of the common programming patterns. I'll be interested to know what
other people think.

One thing to note: In my OpenSCAD2 list library, most functions have
double argument lists, with the list in the second or outmost argument
list. This allows you to chain together list operations the same way you
can chain together geometric transforms in OpenSCAD.

OpenSCAD mailing list
Discuss@lists.openscad.org
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

of the common programming patterns. I'll be interested to know what other
people think.

I think we just don't need such language complexity to describe 3D objects.

It there any 3D object that cannot be described by OpenScad as it stands
now?

I am far more interested in OpenScad producing numerically accurate
results and getting faster than being able to accept more programming
paradigms. Sure some things get expressed more succinctly but at the
expense of being more cryptic and bringing the need to read up on several
languages to understand the code.

I only started this thread to propose a way of not having to create nested
lists just to flatten them. It seems to have lead to a Frankenstein's
monster of a loop syntax that combines bits from everybody's favourite
languages.

Fiddling about with the syntax of the source code when fundamental things
like union and rotate don't produce the correct results seems pointless to
me.

On 7 January 2016 at 17:03, doug moen doug@moens.org wrote:

OpenSCAD mailing list
Discuss@lists.openscad.org
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

On Jan 7, 2016, at 12:40 PM, nop head nop.head@gmail.com wrote:

I think we just don't need such language complexity to describe 3D objects.

It there any 3D object that cannot be described by OpenScad as it stands now?

I am far more interested in OpenScad producing numerically accurate results and getting faster than being able to accept more programming paradigms. Sure some things get expressed more succinctly but at the expense of being more cryptic and bringing the need to read up on several languages to understand the code.

I only started this thread to propose a way of not having to create nested lists just to flatten them. It seems to have lead to a Frankenstein's monster of a loop syntax that combines bits from everybody's favourite languages.

Fiddling about with the syntax of the source code when fundamental things like union and rotate don't produce the correct results seems pointless to me.

On 7 January 2016 at 17:03, doug moen doug@moens.org wrote:
Sorry about the negative tone of my response. I did get frustrated by your original post, though. It's written in a pseudo-code, somewhere between Lisp and OpenSCAD. Since you weren't proposing that as the final syntax, I tried to translate your examples into OpenSCAD syntax, and what I finally ended up with were list comprehensions + Haskell-inspired list functions. Which is not what you were proposing.

The entire Genera OS was written in lisp on a lisp machine. Yes an entire OS in Lisp. It remains the best development system I worked on.

I never had an opportunity to use a Lisp Machine. I'm a student of computer history, though, and it's legendary. I'd like to spend time playing with this software so that I can understand why it's the best development system ever. Maybe somebody will resurrect this in an emulator one day.

• add structured assignment:
  [for ([a b]=[[1 2] [3 4] [5 6]]) [b a] ]
  giving [[2 1] [4 3] [6 5]]

I completely agree with adding structured assignment. I've been writing OpenSCAD2 code as part of a project, and I've added structured assignment, because it is so convenient for destructuring 2-vectors and 3-vectors in the style of geometric programming I'm playing with.

Your example is pseudo code, so I'll translate to OpenSCAD syntax:
[ for ([a,b] = [[1,2], [3,4], [5,6]]) [b,a] ]

In my version of structured assignment,
[a,b] = x
produces a run-time error if x is not a list of length 2. The syntax works in definitions, function formal parameter lists, and for statements.

• add from
  [ for (i=[1:10], j from 2) i*j ]
  where j starts at 2 and counts up. the i variable sets the termination of
  the loop
  from could not appear on its own without a termination clause.

You intend this to mean 2 loops running in parallel, instead of being nested. In OpenSCAD, the syntax
for (i=[1:10], j=[2:11]) ...
looks like it would be good syntax for two parallel loops, but it's not. It actually means nested loops, same as
for (i=[1:10]) for (j=[2:11]) ...
So I think there is a backward compatibility problem with expressing parallel loops in the way you suggest. The problem is in violating the expectations of OpenSCAD programmers, who expect 'for (i=.., j=..)' to mean nested loops.

Your previous post had several examples of parallel loops. Here's one:

• Create pairs
  [for x = [1:5]
  for y = ["A","B","C","D","E"]
  do [x y]  ]
  giving  [[1 A]  [2 B]  [3 C]  [4 D]  [5 E]]

This is more pseudo-code. The problem is that if we convert this into OpenSCAD syntax, we get nested loops:

[for (x = [1:5])
for (y = ["A","B","C","D","E"])
[x,y]]
and the result is a list of 25 pairs.

In functional style programming, you would create the pairs by calling transpose, which in OpenSCAD is matrix transpose from linear algebra.
transpose([[1:5], ["A","B","C","D","E"]])
=> [[1,"A"], [2,"B"], [3,"C"], [4,"D"], [5,"E"]]

(This works in OpenSCAD2 because [1:5] is operationally equivalent to [1,2,3,4,5] in all contexts.)

We could invent a new syntax for parallel loops in OpenSCAD. Maybe this would work:
for (x = xlist | y = ylist) ...

Or we could just use transpose. Because this also works:
for ([x,y] = transpose(xlist, ylist)) ...

• loop terminatoin with while, or until, or just one and use not operator
  for the other - for readability
  [ for (x=1,2,3,4,5,6], while prime(x))  [ x, x*x] ]
  returning list of [x,x^2] until first non-prime encountered.

I'm not sure about the , before the while, for reasons previously discussed. And the argument of 'while' should be parenthesized, for consistency with the rest of OpenSCAD syntax. So maybe this:

 [ for (x=[1,2,3,4,5,6] | while(prime(x))) [x, x*x] ]

In functional programming, you'd implement this using take_while, which returns the longest prefix of a sequence that matches a predicate. In OpenSCAD2, the code would look like this:

 [ for (x=[1,2,3,4,5,6] >> take_while(prime)) [x, x*x] ]

It's not much worse, and no additional syntax is added to the language, since take_while is a library function. There is also a drop_while function which discards initial elements that match the predicate, and I'm not sure if this is expressible using the LOOP macro.

• define more internal loop variables. inner scope only
  [for (x=1,2,3,4,5,6], start=12) x*start ]

But we already have syntax for this.
[for (x=[1,2,3,4,5,6]) let (start=12) x*start ]

• I can't imagine how to do if-then constructs and collect inside the
  existing syntax.

I had the same problem, reading your previous post. You wrote:

• using conditionals within the loop (as an operator, not a regular if
  statement)
  [for x below 10
  if odd(x)
  collect x into odds
  else
  collect x into evens
  finally [odds, evens] ]
  gives [[1 3 5 7 9]  [0 2 4 6 8]]

In functional programming, this would be done using 'partition'.
partition(odd)([0:9])

Here is the non-recursive definition:
partition(f)(s) =
let (p1 = [for (x=s) if (f(x)) x],
p2 = [for (x=s) if (!f(x)) x])
[p1,p2];

In summary, my counter-proposal is to add structured assignment to OpenSCAD2, and then provide library functions to make writing loops and list processing more convenient.

I'll admit I have a bias for the functional style of programming, since I'm used to that, and I don't have any experience using the Lisp LOOP macros. And I admit there's a learning curve; you need a cookbook to learn some of the common programming patterns. I'll be interested to know what other people think.

One thing to note: In my OpenSCAD2 list library, most functions have double argument lists, with the list in the second or outmost argument list. This allows you to chain together list operations the same way you can chain together geometric transforms in OpenSCAD.

OpenSCAD mailing list
Discuss@lists.openscad.org
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

OpenSCAD mailing list
Discuss@lists.openscad.org
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

How do you describe a Moebius ring in OpenScad?

Just curious,
Jean-Paul
N1JPL

of the common programming patterns. I'll be interested to know what other
people think.

objects.

now?

results and getting faster than being able to accept more programming
paradigms. Sure some things get expressed more succinctly but at the
expense of being more cryptic and bringing the need to read up on several
languages to understand the code.

nested lists just to flatten them. It seems to have lead to a
Frankenstein's monster of a loop syntax that combines bits from everybody's
favourite languages.

things like union and rotate don't produce the correct results seems
pointless to me.

your original post, though. It's written in a pseudo-code, somewhere
between Lisp and OpenSCAD. Since you weren't proposing that as the final
syntax, I tried to translate your examples into OpenSCAD syntax, and what I
finally ended up with were list comprehensions + Haskell-inspired list
functions. Which is not what you were proposing.

entire OS in Lisp. It remains the best development system I worked on.

computer history, though, and it's legendary. I'd like to spend time
playing with this software so that I can understand why it's the best
development system ever. Maybe somebody will resurrect this in an emulator
one day.

OpenSCAD2 code as part of a project, and I've added structured assignment,
because it is so convenient for destructuring 2-vectors and 3-vectors in
the style of geometric programming I'm playing with.

works in definitions, function formal parameter lists, and for statements.

of

nested. In OpenSCAD, the syntax

It actually means nested loops, same as

parallel loops in the way you suggest. The problem is in violating the
expectations of OpenSCAD programmers, who expect 'for (i=.., j=..)' to mean
nested loops.

OpenSCAD syntax, we get nested loops:

transpose, which in OpenSCAD is matrix transpose from linear algebra.

[1,2,3,4,5] in all contexts.)

would work:

discussed. And the argument of 'while' should be parenthesized, for
consistency with the rest of OpenSCAD syntax. So maybe this:

returns the longest prefix of a sequence that matches a predicate. In
OpenSCAD2, the code would look like this:

since take_while is a library function. There is also a drop_while function
which discards initial elements that match the predicate, and I'm not sure
if this is expressible using the LOOP macro.

OpenSCAD2, and then provide library functions to make writing loops and
list processing more convenient.

I'm used to that, and I don't have any experience using the Lisp LOOP
macros. And I admit there's a learning curve; you need a cookbook to learn
some of the common programming patterns. I'll be interested to know what
other people think.

double argument lists, with the list in the second or outmost argument
list. This allows you to chain together list operations the same way you
can chain together geometric transforms in OpenSCAD.

OpenSCAD mailing list
Discuss@lists.openscad.org
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

On Jan 7, 2016, at 1:19 PM, nop head nop.head@gmail.com wrote:

http://www.thingiverse.com/thing:239158

On 7 January 2016 at 17:58, Jean-Paul Louis louijp@yahoo.com wrote:
How do you describe a Moebius ring in OpenScad?

Just curious,
Jean-Paul
N1JPL

OpenSCAD mailing list
Discuss@lists.openscad.org
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

OpenSCAD mailing list
Discuss@lists.openscad.org
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org