I was doing a search through an array, and I wanted to assert that I definitely would find what I was looking for in the array.  So I coded this, which crashed DLI 0.1.1:

for(int foo=0; assert(foo<array.length); foo++)
  if(do_test(array[foo]))
    do_stuff();

Is this be legal?  If not, why not?

--
The Villagers are Online! http://villagersonline.com

.[ (the fox.(quick,brown)) jumped.over(the dog.lazy) ]
.[ (a version.of(English).(precise.more)) is(possible) ]
?[ you want.to(help(develop(it))) ]

Russ Lewis wrote:
> I was doing a search through an array, and I wanted to assert that I
> definitely would find what I was looking for in the array.  So I coded
> this, which crashed DLI 0.1.1:
> 
> for(int foo=0; assert(foo<array.length); foo++)
>   if(do_test(array[foo]))
>     do_stuff();
> 
> Is this be legal?  If not, why not?

assert isn't an expression, so no.  This assertion is redundant anyway - in normal builds the array[foo] will do a bounds check, in release builds neither will be checked.

Burton Radons wrote:

> Russ Lewis wrote:
> > I was doing a search through an array, and I wanted to assert that I definitely would find what I was looking for in the array.  So I coded this, which crashed DLI 0.1.1:
> >
> > for(int foo=0; assert(foo<array.length); foo++)
> >   if(do_test(array[foo]))
> >     do_stuff();
> >
> > Is this be legal?  If not, why not?
>
> assert isn't an expression, so no.  This assertion is redundant anyway - in normal builds the array[foo] will do a bounds check, in release builds neither will be checked.

ACK!!!!  I forgot the key thing in the whole code!

There was supposed to be a break if you match the test:

for(int foo=0; assert(foo<array.length); foo++)
  if(do_test(array[foo]))
  {
    do_stuff();
    break;
  }

Does that make more sense now?
--
The Villagers are Online! http://villagersonline.com

.[ (the fox.(quick,brown)) jumped.over(the dog.lazy) ]
.[ (a version.of(English).(precise.more)) is(possible) ]
?[ you want.to(help(develop(it))) ]

"Russ Lewis" <spamhole-2001-07-16@deming-os.org> wrote in message news:3DC2903C.660DA40C@deming-os.org...
> I was doing a search through an array, and I wanted to assert that I definitely would find what I was looking for in the array.  So I coded this, which crashed DLI 0.1.1:
>
> for(int foo=0; assert(foo<array.length); foo++)
>   if(do_test(array[foo]))
>     do_stuff();
>
> Is this be legal?  If not, why not?
>
No its not legal and should never be.
apart from assert not being an expression
assert is a check for something that is NOT expected to happen and should
compile to nothing when non assert builds are built.
here quite clearly foo can be larger than array.length;

I assume your code should have been (where getting to the end would be an
error)
for(int foo=0; assert(foo<array.length); foo++)
   if(do_test(array[foo]))
     break;
arrays should be bounds checked in asserting builds (IMHO)

also asserts SHOULD not have side effects
the code
foo = 0;
do {
    assert( ++foo < array.length );
    count += array[foo];
} while( count < 100 );

should be a compiler warning (modification withing an assert)

Mike.

Russ Lewis wrote:
> Burton Radons wrote:
> 
> 
>>Russ Lewis wrote:
>>
>>>I was doing a search through an array, and I wanted to assert that I
>>>definitely would find what I was looking for in the array.  So I coded
>>>this, which crashed DLI 0.1.1:
>>>
>>>for(int foo=0; assert(foo<array.length); foo++)
>>>  if(do_test(array[foo]))
>>>    do_stuff();
>>>
>>>Is this be legal?  If not, why not?
>>
>>assert isn't an expression, so no.  This assertion is redundant anyway -
>>in normal builds the array[foo] will do a bounds check, in release
>>builds neither will be checked.
> 
> 
> ACK!!!!  I forgot the key thing in the whole code!
> 
> There was supposed to be a break if you match the test:
> 
> for(int foo=0; assert(foo<array.length); foo++)
>   if(do_test(array[foo]))
>   {
>     do_stuff();
>     break;
>   }
> 
> Does that make more sense now?

I understood what you meant.  It turns out assert is an expression - I assume to allow chaining with comma - but it's still incorrect here as it returns void.  I throw an error on the occasion now.

I don't think it's a common enough form to have it in the language.