FW: Ada 200X

FW: Ada 200X

[Erlo Haugen]

-----Original Message-----
From: Gautier Write-only [mailto:gautier@somewhere.nil] 
Sent: 3. juni 2003 04:46
To: comp.lang.ada@ada.eu.org
Subject: Re: Ada 200X


And838N@netscape.net wrote:

> I've heard professors say they don't like Ada because it's slow and 
> does all those "bounds" checking "things".

>They are misinformed: Ada is slow when the bounds checking is ON. After
some debugging phase, if you want, you can turn all checkings OFF ! And as
you point out, better to have> the possibility of bounds checking than
not... ________________________________________________________
Gautier  --  http://www.mysunrise.ch/users/gdm/gsoft.htm

NB: For a direct answer, e-mail address on the Web site!
<


Wouldn't
subtype day_number is natural range 1..31;
type month_days is array (day_number) of some_other_type;

eliminate the boundschecking??

Erlo Haugen

Re: FW: Ada 200X

[Preben Randhol]

Erlo Haugen wrote:
> Wouldn't
> subtype day_number is natural range 1..31;
> type month_days is array (day_number) of some_other_type;
> 
> eliminate the boundschecking??

Not sure what you mean here.

   procedure testing is
      subtype day_number is natural range 1..31;
      type month_days is array (day_number) of Integer;
      Mai : month_days;
   begin
      Mai (32) := 1;
   end testing;

gnatmake testing.adb 
   gnatgcc -c testing.adb
   testing.adb:6:09: warning: value not in range of type "day_number"
   defined at line 2
   testing.adb:6:09: warning: "constraint_error" will be raised at run time
   gnatbind -x testing.ali
   gnatlink testing.ali


./testing 

raised CONSTRAINT_ERROR : testing.adb:6

-- 
Preben Randhol                    http://www.pvv.org/~randhol/

Re: FW: Ada 200X

[John R. Strohm]

"Preben Randhol" <randhol+abuse@pvv.org> wrote in message
news:slrnbdpdan.87p.randhol+abuse@kiuk0152.chembio.ntnu.no...
> Erlo Haugen wrote:
> > Wouldn't
> > subtype day_number is natural range 1..31;
> > type month_days is array (day_number) of some_other_type;
> >
> > eliminate the boundschecking??
>
> Not sure what you mean here.
>
>    procedure testing is
>       subtype day_number is natural range 1..31;
>       type month_days is array (day_number) of Integer;
>       Mai : month_days;
>    begin
>       Mai (32) := 1;
>    end testing;
>
> gnatmake testing.adb
>    gnatgcc -c testing.adb
>    testing.adb:6:09: warning: value not in range of type "day_number"
>    defined at line 2
>    testing.adb:6:09: warning: "constraint_error" will be raised at run
time
>    gnatbind -x testing.ali
>    gnatlink testing.ali
>
>
> ./testing
>
> raised CONSTRAINT_ERROR : testing.adb:6

What he meant is that the naive generated code would be

  if 32 <= 31 then
    Mai(32) := 1;
  else
    raise constraint_error;
  end if;

but then the compiler would notice that 32 <= 31 is a static expression,
evaluate it, and simplify the code to:

  raise constraint_error;

Re: FW: Ada 200X

[Vinzent Hoefler]

Erlo Haugen wrote:

>Wouldn't
>subtype day_number is natural range 1..31;
>type month_days is array (day_number) of some_other_type;
>eliminate the boundschecking??

Inside a function that takes a type day_number as parameter to use it
on an array of type month_days, it definitely should.

The bounds checking should occur *before* calling the function with an
out of range parameter then. Quite hard to determine in C when you
only have ints that can mean everything.


Vinzent.