Ronaldo wrote

I'd strongly suggest not to categorize "if-else" construct as a "filter". A
filter means pick something and discard the rest based on a condition. With
the "if-else", nothing can be discarded (just like cond?a:b), so it
shouldn't be considered a filter. Treating it a filter only add more
confusion 'cos it's the role "if" (w/o else) plays.

Sorry I don't get this part.

Re-arranging "if(a==0) a+1" to "a+(if (a==0)1)" and suggesting they are the
same doesn't make any sense at all. If "a" is subject to the conditional
pick/drop, why would it be placed BEFORE if ? I couldn't ever imagine an
"if" can simultaneously decide the fate of the code before it (a) and after
it (1). That is a mess-up of scopes and only makes it more confusing.

If you ask "why not 'if' shows up here / there if it is allowed in
cond?if(...):if(...)", then with the same logic, since "if" shows up in

why not it is allowed in in

?
To me that's simply a choice of language design.

$  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 ); $ Apps: rollApp , blockscad , openjscad , on AWS ( pdf )

View this message in context: http://forum.openscad.org/Any-where-to-find-the-doc-for-lc-each-lc-else-and-lc-for-c-tp18823p18844.html
Sent from the OpenSCAD mailing list archive at Nabble.com.

runsun wrote

You have answered the question in your preceding message:

runsun wrote

That is exactly what is happening when you write something like:

the filter /if/ is deciding "the fate of the conditional before it". In
fact, the conditional expression above would not act as an expression but as
a filter. And that is confusing!

I agree that is inappropriate to call the construct if-else a filter and
that it is just a conditional that accepts filtering. But it is not
equivalent to ?: because it is not an expression either.

--
View this message in context: http://forum.openscad.org/Any-where-to-find-the-doc-for-lc-each-lc-else-and-lc-for-c-tp18823p18846.html
Sent from the OpenSCAD mailing list archive at Nabble.com.

Ronaldo wrote

My message was to explain that your trivial example is illogical. How could
anyone invent an illogical usage of an idea in order to argue against that
idea ?

/

/

No! it is to decide the fate of the variable AFTER it ... when the condition
"a%2==0" is met, the previous variable of if's, a, is then decided without
invoking the if-filter. So it's not decided by the "if-filter". The 'if'
filter comes into play only when "a%2==0" is not met. Under that condition,
the 'if' filter is to decide the variable/code AFTER it, just like anywhere
'if' is used in this and other languages.

No, the condition expression (the ?:) is not a filter. The "if (a!=1) a*10"
is, working only when condition "a % 2 == 0" is not met.

Re-write it to the currently accepted form:

[ for (a=A) if(a % 2 == 0 ) a else if (a!=1) a*10 ]

Now you have two if's, the 2nd one is a filter but the 1st is not. That's
where the confusion comes.

Yes, "if-else" is a conditional that accepts filtering is the best way to
describe it. Better if added: "if" w/o "else" serves as a filter.

But, other than the filtering, show me an example that can only be expressed
by "if-else" but not "?:".

I don't think you could, because they serve the exact same role (other than
the filtering).

$  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 ); $ Apps: rollApp , blockscad , openjscad , on AWS ( pdf )

View this message in context: http://forum.openscad.org/Any-where-to-find-the-doc-for-lc-each-lc-else-and-lc-for-c-tp18823p18847.html
Sent from the OpenSCAD mailing list archive at Nabble.com.

This will be my third message trying to clarify my point about your proposed
syntax. And, I promise, it will be the last one.

The operator ?: is a ternary operator. As such, it requires 3 operand, no
more no less. In the expression you prefer:

when a=3 the operator ?: receives the following operands: true, 3 and
<nothing>. That is, it receives just 2 operands as the filter if returns
nothing at all. And you expect the operator ?: returns nothing at all in
this case. That is a twisted (and confusing) behavior of it.

The if-else construct on the other hand has the very special property
that it returns nothing if the followed branch (either true or false)
receives nothing. And that is the reason why it cannot be used (like filter*
if*) as a operand of any other operator except itself and the filter* if*.

If we allow the operator ?: receives less that 3 operands why not allow
the sum operator* +* receives less that two?

--
View this message in context: http://forum.openscad.org/Any-where-to-find-the-doc-for-lc-each-lc-else-and-lc-for-c-tp18823p18848.html
Sent from the OpenSCAD mailing list archive at Nabble.com.

Ronaldo, as you write it, ?: is a ternary operator. So it has to return a
result, as it is the outer expression. You seem to argue that this result
may also be nothing, which afaik currently is only an implicit feature (or
call it a hack) of "if".

Ronaldo wrote

I guess, you can always express a filter like this in the right order:
if (a!=1) (a % 2 == 0)? a : a*10

--
View this message in context: http://forum.openscad.org/Any-where-to-find-the-doc-for-lc-each-lc-else-and-lc-for-c-tp18823p18850.html
Sent from the OpenSCAD mailing list archive at Nabble.com.

Parkinbot wrote

I couldn't say it better.

Parkinbot wrote

Surely, we always can but after hard work and at the expenses of clarity in
less obvious cases. The snapshot version syntax solves this and we certainly
can replace all ?: by if-else without fear.

--
View this message in context: http://forum.openscad.org/Any-where-to-find-the-doc-for-lc-each-lc-else-and-lc-for-c-tp18823p18851.html
Sent from the OpenSCAD mailing list archive at Nabble.com.

Parkinbot wrote

This is exactly the same as "if-else" case. The example in the WIP:

could be written /in the right order/, as you wish, as :

and they return the same thing, which once again implies that "?:" behaves
exactly like "if-else", which is what I've been arguing all along.

Parkinbot wrote

I believe that this belief of what "if-else" is is the source of conflict
that causes conversations between Ronaldo and I went parallel.

Afaik, "if-else" by itself has to return something for every single item it
handles, therefor is a "catch-all" construct and shouldn't be considered a
filter. Ronaldo, on the other hand, thinks "if-else" is a filter, that it
"has the very special property that it returns nothing if the followed
branch (either true or false) receives nothing," therefor it can be used
with "if(w/o else)" filter, but "?:" cannot.

Ronaldo forgot one thing: the "special property" he mentioned is a result
of our design, NOT the reason for our design. One cannot design a behavior,
then turn around to argue that having that behavior is the reason we need to
design that behavior, or even go further to argue that we cannot add that
same behavior to others because they don't have it.

"if-else" by itself is never a filter. Even in OpenSCAD when we allow
"if(w/o else)" to be in "if-else" branch, it's the "if(w/o else)" that
executes the filtering, not the "if-else". The ability of  "if-else" to
accept "if(w/o else)" and turn its branch into a filter is NOT a nature of
"if-else", but an add-on feature we intentionally put it in.

In terms of language feature, having a filter in a conditional expression is
a novel and welcomed idea. But mixing "if-else" and "if(w/o else)" means
that users have to deal with that an "if" in OpenSCAD list comprehension is
sometimes a filter, sometimes not. That's where the confusion is.

$  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 ); $ Apps: rollApp , blockscad , openjscad , on AWS ( pdf )

View this message in context: http://forum.openscad.org/Any-where-to-find-the-doc-for-lc-each-lc-else-and-lc-for-c-tp18823p18856.html
Sent from the OpenSCAD mailing list archive at Nabble.com.

I must admit that the idea of having if/else, if within an expression
having this semantics is a bit awkward. Personally I wouldn't introduce it,
but implement the filter by introducing a new symbol like null that
explicates what happens:

in lc-context null could resolve to nothing and in expression context to 0
(or maybe undef).
In the second step I'd define a binary operator op1?op2 that defaults to
the ternary operator op1?op2:null. This new binary operator can then be
used anywhere.

--
View this message in context: http://forum.openscad.org/Any-where-to-find-the-doc-for-lc-each-lc-else-and-lc-for-c-tp18823p18857.html
Sent from the OpenSCAD mailing list archive at Nabble.com.

As I said earlier: in conjunction with each if / else can add different
numbers of elements to a list, which ? : can't as it is a single expression.

E.g.
if(a %2) a else each a, a * a

On 1 November 2016 at 19:17, Parkinbot rudolf@parkinbot.com wrote: