The current snapshot of OpenSCAD supports let, echo and assert in
expressions. The latter two are experimental features. I want to provide
some feedback about those features before the release.

I'm not happy about the different coding style that you have to use for
writing functions vs writing modules. I don't like the lack of consistency.
Also, the module idioms are better, easier to use than the idioms we are
now introducing for writing functions. I'd like to suggest an alternative.

We'll start with local definitions. In a module, you introduce local
definitions like this:
module m(x) {
a = ...;
b = ...;
code...
}
This used to be done using assign, but that's deprecated because the new
syntax is better.

In a function, you introduce local definitions like this:
function f(x) =
let (a = ...,
b = ...)
code;

The function idiom is basically the same as assign, which was deprecated
in favour of the new better syntax. I don't have a problem with let, it
is actually convenient and looks good in some contexts, but it's
inconsistent that you are forced to write modules and functions
differently, and that bugs me.

Now consider echo and assert. The syntax is again different between
modules and functions. That's because, in an expression, echo and
assert have an extra trailing argument which is the expression to
evaluate once the action has been taken.

module m(x) {
echo("m: x=",x);
a = ...;
b = ...;
code...
}
function f(x) =
echo("f: x=",x)
let (a = ...,
b = ...)
code;

Now suppose that I want a conditional echo. I don't want to trace every
call to the module/function, only certain special calls for some edge
condition.

For the module, that's easy:
module m(x) {
if (x < 0) echo("m: x=",x);
a = ...;
b = ...;
code...
}

For the function, it's possible, but it's an intellectual challenge to
figure out how to do it, and it's definitely ugly. This is the point where
the inconsistency between modules and functions blows up and becomes ugly
for me.

So I have an alternative suggestion. I want to introduce a new kind of
expression, called a 'qualified expression'. It consists of

• '('
• a sequence of one or more 'statements', each terminated by ';'.
• an expression, which is evaluated to produce the result
• ')'

A 'statement', in this context, is either a definition (like 'x=0') or an
action (like echo() or assert()). An action can be made conditional, or you
can put an action in the body of a 'for' loop. (In my implementation, I'm
actually reusing the code for 'if' and 'for' that's used in list
comprehensions, because it's the same syntax with just a different
interpretation.)

Here's how to write the function with a conditional echo, using this new
syntax:

function f(x) = (
if (x<0) echo("f: x=",x);
a = ...;
b = ...;
code
);

I'm interested if other people also think this is a good idea. I'm starting
to implement this in my OpenSCAD2 prototype. I'm far enough along that I
know this idea is coherent and implementable.

btw where is the assert() ? Produce "unknown module assert()" in the snapshot
I am using (2016.10.04 (git 7e0935d)), which I suppose is current ?

$  Runsun Pan, PhD $ libs: doctest , faces ( git ), offline doc ( git ), runscad.py ( 2 , git ), synwrite ( 2 );   $ tips: hash ( 2 ), matrix ( 2 , 3 ), sweep ( 2 , 3 ), var ( 2 ), lerp , animation ( gif , prodVid , animlib ), precision ( 2 ), xl-control , type , rounded polygon , chfont , tailRecur ( 2, 3 ), isosphere ( 2 ), area , vol/center , RGB , CurvedImg , tests ( 2 ), text ; $ Apps: rollApp , blockscad , openjscad , on AWS ( pdf )

View this message in context: http://forum.openscad.org/feedback-let-echo-and-assert-in-expressions-tp19111p19113.html
Sent from the OpenSCAD mailing list archive at Nabble.com.

I see. If we go this way, we will have 3 different constructs for "if-else":

1. In top-level and module:

   if(...) some_code_line;
   if(...){ } else { }

2. In function:

   if(...) some_code_line;
   if(...)( ) else ( );

3. In list comprehension:

   [ for(...) if(A) X ]
   [ for(...) if(A) X else Y ]
   [ for(...) if(A) if(B) X else Y ]
   [ for(...) if(A) ( if(B) X else Y ) ]
   [ for(...) if(A) ( if(B) X) else Y ]
   [ for(...) if(A) X else if(B) Y ]

Not sure about this, man. The inspiration of this change is that

"it's inconsistent that you are forced to write modules and functions
differently"

The intended change would mean that, in an attempt to solve that
inconsistency, we will create another inconsistency, while we are still
"forced to write modules and functions differently" (i.e., the problem isn't
actually solved), and this is on top of what already confusing usage of
if-else in the list comprehension.

$  Runsun Pan, PhD $ libs: doctest , faces ( git ), offline doc ( git ), runscad.py ( 2 , git ), synwrite ( 2 );   $ tips: hash ( 2 ), matrix ( 2 , 3 ), sweep ( 2 , 3 ), var ( 2 ), lerp , animation ( gif , prodVid , animlib ), precision ( 2 ), xl-control , type , rounded polygon , chfont , tailRecur ( 2, 3 ), isosphere ( 2 ), area , vol/center , RGB , CurvedImg , tests ( 2 ), text ; $ Apps: rollApp , blockscad , openjscad , on AWS ( pdf )

View this message in context: http://forum.openscad.org/feedback-let-echo-and-assert-in-expressions-tp19111p19117.html
Sent from the OpenSCAD mailing list archive at Nabble.com.