PRIMO STRAW POLL / New Op!'s

Question:
Should paren!'s be necessary to enforce
correct evaluation order in PRIMO when
using infix operators?


examples of preferable behaviour?

>> a: 4
== 4
>> print (a + 2)
6
== unset
>> a:= (a + 2)
== 6
>> a
== 6
>> a:= add a 2
== 8
>> a
== 8


however is this correct behaviour?

>> a: 4 ; this part is obviously okay
== 4
>> print a + 2
4
** Script error: + expected val1 of type number! not unset!
** near: print a + 2
>> a:= a + 2 ; reset 'a to :a THEN perform addition?
== 6
>> a
== 4
>> a:= add a 2
== 6
>> a
== 6

Let me know what you all think about this.

Ladislav, I do like your idea very much for using
"word::" as syntax for reset-word! types and it does
type much easier, so Iam considering it deeply, however
Iam sticking to my original idea of := for these
examples, I also think := is more visually distinctive
on page / screen than :: and has some historical
precedence from other languages, but I DO like the
ergonomic properties of your suggestion so it is still
up in the air as far as final syntax goes, but.....

I've been using a bit of PIKE recently and it's great
for prototyping C, C++ type programs in an interpretative
manner and it has properties which enable me to play
about with ideas for series! block! arrays vectors etc.
for PRIMO as well as infix operators.

What do you think of additional infix op! functions that
have destructive binding properties like reset-word!'s do.
Iam think of the C like operators that C, C++, Perl, Python
and Pike all share, here's what Iam proposing,

>> a: 4
== 4
>> a += 5
== 9
>> a
== 9
>> a:= 0
== 0
>> a
== 0
>> a:= (a + 1)
== 1
>> preset-add a 5
== 6
>> a
== 6
>> a += 5 ; Infix op! += equivalent to preset-add word value.
== 11
>> a
== 11
>> postset-add a 5
== 11
>> a
== 16
>> a =+ 5 ; Infix op! =+ equivalent to postset-add word value.
== 16
>> a
== 21
>> a:= unset ; same as to-unset a
== unset
>> a
** Script error: word a is an unset word, define word before use
** near: a
>> unset-word? a ; no need to quote word as func arg is a quoted 'value
== true

Viewed in this light, if these op!'s are acceptable then word:=
for reset-word! seems more in keeping with reset operators like
+= , -= , =+ etc.

Here's a full list of proposed op!'s with optional alternatives in
brackets to help ease people coming from other languages.

current op!'s

+ - * / ** (^) // (%) >= > = == < <= <> (!=) *see note below
and or xor

new op!'s

preset types perform action then return value

+= -= *= /= //= **= (^=) (%=)

postset types return existing word value then reset word
after peforming action.

=+ =- =* =/ =** =// (=^) (=%)

Interesting that the != for inequality is still lurking in REBOL
as anassignable infix op! value, see below.
>> !=: get to-word "<>"
>> 4 != 5
== true
>>

All the proposed new op!'s have their equivalent prefix functions.
+= or preset-add
=+ or postset-add
**= or preset-power
=** or postset-power

Might as well thrown in

Increment [value [word! number!]]
Decrement [value [word! number!]]

increment same as word += 1 or preset-add word 1
decrement same as word -= 1 or preset-subtract 1

note in these examples the :word must be a number!
for these funcs to work properly.

Anybody any comments?

cheers,

Mark Dickson