[fricas-devel] Unexpected behavior of commas in singleton blocks {expression}

[Riccardo GUIDA]

Hi!

I bet that the behavior I describe below is well known to fricas developers,
but I report this because I find it fun and awful at the same time...

1) WARM-UP:
IF I understand well, my UG's readings, SPAD and Aldor have **different semantics** for commas. N>=2 below.

* in Aldor (expr1, expr2,..,exprN) is a tuple (multivalue) and {expr1; expr2;..; exprN}
is a sequence of expression evaluated in that order. Parentheses () and braces {} are
used for precedence's sake and 99% interchangeable (different behavior for external trailing semicolon).
Aldor allows internal trailing semicolon in blocks {e1;e2;e3;} while this is an error in spad.

* In Spad (expr1, expr2,..,exprN) is a tuple (not really, see point 2)
while {e1;..; eN} **AND** {e1,..,eN} are sequences of expressions evaluated in the order.

OK, I find Aldor semantics much more clean & logic than that of Spad, but this is not the problem.

2) THE MYSTERY:
The mysterious behavior is this one: one would expect that "{expression}" has a value equal to "expression,
but this is NOT the case in the interpreter if "expression" is a tuple.

(1) -> {1;2;3} -- this is a sequence of expressions in a block, OK.

(1) 3
Type: PositiveInteger
(2) -> {1,2,3} -- this is a sequence of expressions in a block, OK.

(2) 3
Type: PositiveInteger
(3) -> 1,2,3 -- tuple

(3) [1, 2, 3]
Type: Tuple(PositiveInteger)
(4) -> (1,2,3) -- tuple

(4) [1, 2, 3]
Type: Tuple(PositiveInteger)
(5) -> {"bum"} --{expr} same as expr? YES

(5) "bum"
Type: String
(6) -> {(1,2,3)} --{expr} same as expr? NOOOO

(6) 3
Type: PositiveInteger
(7) -> ((1,2),(3,4)) --tuple of tuples

(7) [(1, 2), (3, 4)]
Type: Tuple(Tuple(PositiveInteger))
(8) -> {((1,2),(3,4))} -- tuple of tuples? NOOOOO

(8) [3, 4]
Type: Tuple(PositiveInteger)

In the example above I would have expected [1,2,3] as result of (6) and [[1,2],[3,4]] as result of (8).

As a second choice I would have expected syntax errors because, for instance,
in (6) I try to see "(1" as first expression in a comma separated sequence
of expressions but it is not a well formed expression. Same for "((1,2)" in (8).

It is like the semantics of commas (,...,) for tuple is forgotten inside
a singleton block and mutated to that of a sequence.

weird feature...

regards
ric

[Ralf Hemmecke]

Post by Riccardo GUIDA
weird feature...

I have no idea how to explain this behaviour, but it is great that you
report such "funny" functionality.

I hope, some day, they all will be replaced by a clear (and simple)
semantics.

Ralf