A = A + 1;

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

oops

"The thing to remember with OpenSCAD is that it is timeless."

This cannot be true.  Otherwise, "rotate translate object" would be non-deterministicly identical to "translate rotate object", and they're not.  Where object ends up depends on which transformation gets executed first.  And if something gets executed "first", then time is involved.

On Tue, Feb 3, 2015 at 4:28 PM, Michele <[hidden email]> wrote:
> /"The thing to remember with OpenSCAD is that it is timeless."/
>
> This cannot be true.  Otherwise, "rotate translate object" would be
> non-deterministicly identical to "translate rotate object", and they're not.
> Where object ends up depends on which transformation gets executed first.
> And if something gets executed "first", then time is involved.

"Timeless" does not mean "Non-Hierarchical".  R T O is not the same as
T R O because it's R->T not T->R.

There is an order, but it's not iterative nor is there a frozen
"moment in time" one could peek at the state of things and alter the
outcome.

-ethan

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

I don't have a problem with OpenSCAD, because I treat it as a functional language (with some recursion limitations), and everything is fine. There are ways to implement variable modification in functional languages by including a data store as an additional function parameter -- I'd have to dig out my CS notes to find the way I was taught -- but it's frequently the case that such a design is not necessary.

The 'concat' operator was, in my opinion, the final advance to make the language essentially turing complete (and therefore fully expressive). However, for proper use of functional languages, recursion is required in many cases.

Thinking about the vector sum question, here's a recursive version that basically uses assignment. The "cumSum" argument stores the current sum, and the "pos" argument stores the current location in the vector:

function vsum(vec, pos = 0, cumSum = 0) =
  pos >= len(vec) ? cumSum : vsum(vec, pos+1, cumSum + vec[pos]);

echo(vsum([1,2,3,8]));

I see three complaints.

* OpenSCADs handling of variables is suboptimal. True, although there were some large improvements recently.

* OpenSCAD is lacking as a functional language. True.

* OpenSCAD lacks a destructive update operator (c-style assignment). This is inherent to the way Openscad models the target domain (constructive solid geometry) and it can not possibly change, because it would violate that union is a commutative operator.
A destructive update operator would have to be valid only in certain very limited places, causing even more confusion.

Example:

b = 5;

union() { // Union operator
   
    // Operand A
    union() {
        b = b + 5;
        sphere(d=b, $fn=50);
    }
   
    // Operand B
    union() {
        b = b + 5;
        translate([40, 0, 0])
        cube(b, center=true);
    }
}

Both shapes have size 10, because each sub-union() creates its own variable b that shadows the global b. In C you would be able to re-assign the global b. In this case the sphere would have d=5+5=10, and the cube would get size 5+5+5 = 15.
By swapping the order of operand a and b, the sphere would be 15 and the cube 10.

Union() is a commutative operator, meaning that the result does not depend on the order of the operands. Allowing reassignment breaks this fundamental property of union().

Have a nice day,
Bananapeel :)

MichaelAtOz wrote

I really think we should use 'constant' or some equally descriptive term. (define value? like #define)?

Bite the bullet, this keeps coming up so often...

But we need to clarify the iterative list comprehension/recursion in expressions, v's the for() loops/if() in geometry... I think we are having/approaching a clash of terminology of module geometry instantiation v's expression evaluation.

Let the battle begin....

Yes we have been over this before more than once.

Once you grok OpenSCAD it all makes sense.

Unless I see objections, I will start a project to change the wiki to use 'identifier', 'definition', 'constant' and 're-definable' as needed, and point out the -D rationale. [eventually...]

MichaelAtOz wrote

Unless I see objections, I will start a project to change the wiki to use 'identifier', 'definition', 'constant' and 're-definable' as needed, and point out the -D rationale. [eventually...]

You have my objection. They are variables in the mathematical sense. A symbol that represents a quantity in a mathematical expression.

http://en.wikipedia.org/wiki/Variable
Variable (mathematics), a symbol that represents a quantity in a mathematical expression, as used in many sciences
Variable (computer science), a symbolic name associated with a value and whose associated value may be changed

I'm comfortable with calling them variables, but I suppose they could be called parameters instead. The wikipedia definition suggests that "parameter" might be appropriate in a mathematical sense:

http://en.wikipedia.org/wiki/Parameter#Mathematical_functions

"Mathematical functions have one or more arguments that are designated in the definition by variables. A function definition can also contain parameters, but unlike variables, parameters are not listed among the arguments that the function takes. When parameters are present, the definition actually defines a whole family of functions, one for every valid set of values of the parameters."

The Variable wikipedia page defines parameters as a subclass of variables:

http://en.wikipedia.org/wiki/Variable_%28mathematics%29#Specific_kinds_of_variables

"To distinguish them, the variable x is called a unknown, and the other variables are called parameters or coefficients, or sometimes constants, although this last terminology is incorrect for an equation and should be reserved for the function defined by the left-hand side of this equation."

It also works well with calling OpenSCAD files "parametric designs". However, "parameters" and "unknowns" are indistinguishable in code (and in a formula, for that matter), which would make explaining the difference a little bit more difficult to a naive person: unknowns appear in the argument list of the function.

The "constants" that are used in OpenSCAD cannot be changed, so they are
not "variable" or "variables".

These discussions about "variables" come up once a month.  They consume
time. And they frustrate many people away from OpenSCAD (I assume that
for every newbie who persists in asking questions here, a dozen simply
give up and wander off).

In my opinion, the OpenSCAD community needs to explain the language in a
way that is accessible to the average programmer (think "C" rather than
"declarative") without many newbies having to re-invent the variable
wheel.  The discussion around OpenSCAD " variables" in the
manual/documentation needs to include the meat of the discussions that
we have on a regular basis with newbies.

Jon

On 2/3/2015 6:36 PM, Bananapeel wrote:

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

"You have my objection. They are variables in the mathematical sense. A
symbol that represents a quantity in a mathematical expression."

I agree with Mr. Peel and cast my vote similarly.

Yes, but using  the mathematical definition contributed to this problem.

To the layperson variable usually implies, in this context,  a programming variable. While that person is also wrong, I think it is pragmatic to avoid confusion in the users if it is inexpensive to do so.

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

I disagree with your objections. If they are variables in mathemarical sense, they are also constants in mathematical sense. Since they are constants in programing sense too, i see broader apeal in simply calling them constants and avoid controvercy.

-------- Original message --------
From: Michele <[hidden email]>
Date:03/02/2015 19:59 (GMT-05:00)
To: [hidden email]
Cc:
Subject: Re: [OpenSCAD] A = A + 1;

/"You have my objection. They are variables in the mathematical sense. A
symbol that represents a quantity in a mathematical expression."/

I agree with Mr. Peel and cast my vote similarly.



--
View this message in context: http://forum.openscad.org/A-A-1-tp11385p11429.html
Sent from the OpenSCAD mailing list archive at Nabble.com.

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

-------- Original message --------
From: "David Eccles (gringer)" <[hidden email]>
Date:03/02/2015 17:24 (GMT-05:00)
To: [hidden email]
Cc:
Subject: Re: [OpenSCAD] A = A + 1;

I don't have a problem with OpenSCAD, because I treat it as a functional
language (with some recursion limitations), and everything is fine. There
are ways to implement variable modification in functional languages by
including a data store as an additional function parameter -- I'd have to
dig out my CS notes to find the way I was taught -- but it's frequently the
case that such a design is not necessary.

The 'concat' operator was, in my opinion, the final advance to make the
language essentially turing complete (and therefore fully expressive).
However, for proper use of functional languages, recursion is required in
many cases.

Thinking about the vector sum question, here's a recursive version that
basically uses assignment. The "cumSum" argument stores the current sum, and
the "pos" argument stores the current location in the vector:

function vsum(vec, pos = 0, cumSum = 0) =
  pos >= len(vec) ? cumSum : vsum(vec, pos+1, cumSum + vec[pos]);

echo(vsum([1,2,3,8]));



--
View this message in context: http://forum.openscad.org/A-A-1-tp11385p11422.html
Sent from the OpenSCAD mailing list archive at Nabble.com.

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

adriang wrote

Does the following
echo(vsum([[1,2],[3,4]]));
work for your definition?

I don't know any normal language that would automatically allow vsum([1,2,3,8]) and vsum([[1,2],[3,4]]) into the same function and give a sensible result. They have different types. One is a vector of integers, the other is a vector of vector or integers. That's like applying an integer function to a string and expecting it to magically figure out what you want.

Would you expect the result to be 10 or [4, 6]?

Every other language without destructive update operator (c-style assign) calls them variables, I don't see why OpenSCAD should be different.

They aren't "constant", because a constant is a value that never changes.
function square(s) = s*s;
b = 6;
a = square(b);
The variable a's value isn't constant - it changes according to the value of b.

What's really needed is:
* Giving an error message when trying to redefine a variable.
* Clean up other doubtful cases.
* An explanation for people from procedural programming languages (C-style).
* Push all these changes to release version.

Have a nice day,
Bananapeel :)

Hmm, I am no expert, but in your example below, 'a' is not varying, it
is constant - it gets calculated once and set up for use.

It would only be a variable if it varies during the program run, eg if
'b' changes at certain points in time....but openscad never 'runs' it
just puts the STL together, and 'b' is always the same.

On 04/02/2015 13:03, Bananapeel wrote:

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

The main page of the manual: http://en.wikibooks.org/wiki/OpenSCAD_User_Manual
refers to an openscad source file as a script. The word script makes it sound iterative to me. Is there a better way to explain it?

"Anonymous scopes are not considered scopes":
{
    angle = 45;
}

IMO this should be a separate scope, or give an error.