I think d should add an operator for assignment ":=" instead of "==" like Perl 6. http://dev.perl.org/perl6/doc/design/apo/A03.html

I recently had trouble with == and = and I think D should add a Pascal like equals operator. This would be an improvement and allow the avoidance of a == d + 1; errors

ie if (a == d) ...
ie if (a := d) ...

:=

Urza a écrit :
> I think d should add an operator for assignment ":=" instead
> of "==" like Perl 6.
> http://dev.perl.org/perl6/doc/design/apo/A03.html

Uh, could you explain better what's the difference between = and := ?

The explanation in the Perl webpage is .. weird .. (that said being a webpage about Perl, I'm not surprised).


> I recently had trouble with == and = and I think D should add a
> Pascal like equals operator. This would be an improvement and allow
> the avoidance of a == d + 1; errors
> 
> ie if (a == d) ...
> ie if (a := d) ...
> 
> :=

A bad idea: = is shorter than :=, less character: less cruft.
The 'correct' way to avoid the == and = confusion is to make if accept only booleans: this way if a and d are int 'if (a = d)' is incorrect.

The only remaining issue is that it doesn't totally avoid the problem: you can still have = and == confusion when a and d are boolean, an error quite hard to find..

Another solution is the gcc hack of making the compiler warn on any = inside a 'if' except when there is a (( )), it's a hack but it's a more complete solution..


That said, given that D is based on C/C++ syntax (it even reused C's ugly syntax for variable declaration instead of using a nice syntax like Limbo or Scala do), I doubt that you'll get much success for  the replacement of = by :=.

renoX

renoX wrote:

> Another solution is the gcc hack of making the compiler warn on any =
> inside a 'if' except when there is a (( )), it's a hack but it's a more
> complete solution..

When should an assignment be used as part of an expression anyway? It's only a way to make code more compact and less readable.

Maybe an assignment just shouldn't return anything.

I think the compiler should still be able to implicitly cast int to bool, though.

-- 
Michiel

Heehee... today I retired from my job.
On thursday I'm going to a tropical island.  That said...


I think the real solution is to just use the right operator, and recognize the difference between = and ==, just like you know to put a semicolon at the end.  It's trivial.

Worry about metaprogramming or 'in' operator for normal arrays, proveability or something.  : p

Michiel Helvensteijn Wrote:
> renoX wrote:
> 
> > Another solution is the gcc hack of making the compiler warn on any =
> > inside a 'if' except when there is a (( )), it's a hack but it's a more
> > complete solution..
> 
> When should an assignment be used as part of an expression anyway? It's only a way to make code more compact and less readable.
> 
> Maybe an assignment just shouldn't return anything.
> 
> I think the compiler should still be able to implicitly cast int to bool, though.
> 
> -- 
> Michiel
>