[Aldor-l] with issue

[Ralf Hemmecke]

On 07/05/2008 05:23 PM, Peter Broadbery wrote:
> On Sat, Jul 5, 2008 at 10:58 AM, Ralf Hemmecke <ralf at hemmecke.de> wrote:
>> Hi Peter, hi Stephen
>>
>> I remember that long time ago I was told that an instance where
>>
>>  A with B
>>
>> behaves different from
>>
>>  with {A; B}
>>
>> is a bug.
>>
>> What would you say about the following?
>>
>> ---BEGIN aaa.as
>> #include "aldor"
>> A: with == add;
>> Cat: Category == with;
>> Cat2: Category == with;
>> extend IntegerSegment(Z: IntegerType) : with {
>>        Cat with {Cat2; foo: A -> A};
>> } == add {foo(a: A): A == a;}
>> ---END aaa.as
>>
>> Compiling with
>>
>>  aldor aaa.as
>>
>> results in
>>
>> "aaa.as", line 5: extend IntegerSegment(Z: IntegerType) : with {
>>                  .............................................^
>> [L5 C46] #1 (Error) Improper form appearing in `with' statement.
>>
>> Well, of course in that case it is easy to fix the sources, but the question
>> is whether the above program should be considered a well-formed Aldor
>> program. Is it?
>>
>> Ralf

> I don't think '(with C) with X' work gets through the parser either
> (not in a position to check it at the moment), so it's consistent.
> Given the parser refuses to look at the first one, making them both
> work might well be a problem.

Ooops, I would have thought, my code parses as

(*)   with (Cat with {Cat2; foo: A -> A})

instead of

(+)   (with Cat) with {Cat2; foo: A -> A}

as you see it. Why are the braces for, then?
Shouldn't they group together a sequence of statements and consider them 
basically as one statement? So (*) should apply and not (+). Where is my 
thinking error?

Ralf