Put the access value

Put the access value

[ldries46]

I have a situation in which it is not practical to use the debugging  
possibilities of GNAT GPS. Instead I want to use the Put procedure to 
show certain values.
Normally that is no such a program with the Predifined Language 
attributes (mostly with the 'image attribute).
Now I have the following:

type Buffer_Pointer is limited private;

    type Block_Buffer is record
       nr       : integer;
       buf      : Item;
       previous : Buffer_Pointer := null;
       next     : Buffer_Pointer := null;
    end record;

       El  : Buffer_Pointer := LastBuffer;
       El1 : Buffer_Pointer;

I just want to see if the routing of thedifferent Buffer_Pointer's is 
correct so I thought Buffer_Pointer'Image(El) would show the value of 
The Pointer El f.i. ?x000000 for null or even the simpel decimal value 0.

This construction creates a failure during compiling.
Should I use another attribute or some other construction?

Re: Put the access value

[J-P. Rosen]

Le 14/04/2020 à 09:15, ldries46 a écrit :
> I just want to see if the routing of thedifferent Buffer_Pointer's is
> correct so I thought Buffer_Pointer'Image(El) would show the value of
> The Pointer El f.i. ?x000000 for null or even the simpel decimal value 0.
> 
> This construction creates a failure during compiling.
> Should I use another attribute or some other construction?
In Ada, a pointer is not an integer and has no 'Image attribute. A
pointer is not an address either. Of course, for debugging you can
indulge your self to constructs that would be thrown at for long term
maintenance. So...

1) Use Unchecked_Conversion to convert it to an appropriate integer type

2) use package Address_To_Access conversion to convert your pointer to
an address, then System.Storage_Elements.To_Integer to convert the
address to Integer_Address, which is an integer type.

-- 
J-P. Rosen
Adalog
2 rue du Docteur Lombard, 92441 Issy-les-Moulineaux CEDEX
Tel: +33 1 45 29 21 52, Fax: +33 1 45 29 25 00
http://www.adalog.fr

Re: Put the access value

[Oliver Kellogg]

On Tuesday, April 14, 2020 at 9:42:56 AM UTC+2, J-P. Rosen wrote:
> Le 14/04/2020 à 09:15, ldries46 a écrit :
> > [...]
> In Ada, a pointer is not an integer and has no 'Image attribute. A
> pointer is not an address either. Of course, for debugging you can
> indulge your self to constructs that would be thrown at for long term
> maintenance. So...
> 
> 1) Use Unchecked_Conversion to convert it to an appropriate integer type
> 
> 2) use package Address_To_Access conversion to convert your pointer to
> an address, then System.Storage_Elements.To_Integer to convert the
> address to Integer_Address, which is an integer type.

There's also the sledgehammer-to-crack-a-nut solution:

https://github.com/persan/auto-io-gen

- Oliver

Re: Put the access value

[Robert A Duff]

"J-P. Rosen" <rosen@adalog.fr> writes:

> In Ada, a pointer is not an integer and has no 'Image attribute.

Sure it does.  ;-)

So do records and everything else.  See AI12-0020-1 (don't pay attention
to details; they're changing).  I implemented that recently, so the
latest development version of GNAT has it.

This program:

with Ada.Strings.Text_Output.Formatting; use Ada.Strings.Text_Output;
procedure Access_Image is
   type R is record
      This : Integer;
      That : String (1..10);
   end record;

   type A is access all R;

   X : A := new R'(This => 123, That => "helloworld");
begin
   Formatting.Put ("\1, \2\n", X'Image, X.all'Image);
end Access_Image;

prints:

(access 162b740),
(this =>  123,
 that => "helloworld")

> 1) Use Unchecked_Conversion to convert it to an appropriate integer type
>
> 2) use package Address_To_Access conversion to convert your pointer to
> an address, then System.Storage_Elements.To_Integer to convert the
> address to Integer_Address, which is an integer type.

Right, these are good workarounds if you don't have the latest
version.

- Bob

Re: Put the access value

[briot.emmanuel]

> > In Ada, a pointer is not an integer and has no 'Image attribute.
> 
> Sure it does.  ;-)


Hey, that's GNAT specific again, no other compiler implements that yet ! :-)

This is very nice, indeed. We (and I would guess a lot of others) have come up with some code-generation scheme to generate those output routines for our types.
We use libadalang for now (distributed with GNAT). Although our home-made solution
is mode flexible (with binary or text output, for instance, as well as backward compatibility with older type definitions), having this service part of the compiler definitely removes some of the needs for custom frameworks...

Some of the difficulties I had in implementing the code generation was for generics and their instances, or class-wide types. Are those handled natively
by the compiler now, when using 'Image ?


Thanks for letting us know Bob

Re: Put the access value

[Robert A Duff]

briot.emmanuel@gmail.com writes:

> Some of the difficulties I had in implementing the code generation was
> for generics and their instances, or class-wide types. Are those handled
> natively
> by the compiler now, when using 'Image ?

Put_Image works for types in instances of generics.
It does not work for class-wide types -- still working on
that.

> Thanks for letting us know Bob

You're welcome.

- Bob

Re: Put the access value

[Jeffrey R. Carter]

On 4/14/20 9:15 AM, ldries46 wrote:
> 
> type Buffer_Pointer is limited private;
> 
>     type Block_Buffer is record
>        nr       : integer;
>        buf      : Item;
>        previous : Buffer_Pointer := null;
>        next     : Buffer_Pointer := null;
>     end record;
> 
>        El  : Buffer_Pointer := LastBuffer;
>        El1 : Buffer_Pointer;
> 
> I just want to see if the routing of thedifferent Buffer_Pointer's is correct so 
> I thought Buffer_Pointer'Image(El) would show the value of The Pointer El f.i. 
> ?x000000 for null or even the simpel decimal value 0.

1. There are no access types in this code. The declaration of type Block_Buffer 
is invalid because null cannot be a valid visible value of type Buffer_Pointer

II. Assuming the full type of Buffer_Pointer is an access type, and the 
declaration of Block_Buffer can see the full type, it appears you are creating a 
linked list. Why not use Ada.Containers.Doubly_Linked_Lists?

C. Assuming you're still going to use access types, why are you interested in 
the internal representation of access values? These will probably appear to be 
random values that provide no information, except perhaps whether the value is null

iv. If you're only interested in whether an access value is null or not, this 
can be better determined without showing the internal representation:

"El is null " & Boolean'Image (El = null)

"El is " & (if El = null then "" else "not ") & "null"

-- 
Jeff Carter
"Your mother was a hamster and your father smelt of elderberries."
Monty Python & the Holy Grail
06

Re: Put the access value

[ldries46]

Op 14-4-2020 om 13:05 schreef Jeffrey R. Carter:
> On 4/14/20 9:15 AM, ldries46 wrote:
>>
>> type Buffer_Pointer is limited private;
>>
>>     type Block_Buffer is record
>>        nr       : integer;
>>        buf      : Item;
>>        previous : Buffer_Pointer := null;
>>        next     : Buffer_Pointer := null;
>>     end record;
>>
>>        El  : Buffer_Pointer := LastBuffer;
>>        El1 : Buffer_Pointer;
>>
>> I just want to see if the routing of thedifferent Buffer_Pointer's is 
>> correct so I thought Buffer_Pointer'Image(El) would show the value of 
>> The Pointer El f.i. ?x000000 for null or even the simpel decimal 
>> value 0.
>
> 1. There are no access types in this code. The declaration of type 
> Block_Buffer is invalid because null cannot be a valid visible value 
> of type Buffer_Pointer
>
> II. Assuming the full type of Buffer_Pointer is an access type, and 
> the declaration of Block_Buffer can see the full type, it appears you 
> are creating a linked list. Why not use 
> Ada.Containers.Doubly_Linked_Lists?
>
> C. Assuming you're still going to use access types, why are you 
> interested in the internal representation of access values? These will 
> probably appear to be random values that provide no information, 
> except perhaps whether the value is null
>
> iv. If you're only interested in whether an access value is null or 
> not, this can be better determined without showing the internal 
> representation:
>
> "El is null " & Boolean'Image (El = null)
>
> "El is " & (if El = null then "" else "not ") & "null"
>
Thanks to mr J.P Rosen this worked
and I have checked a Buffer random Buffer insert routinethat looked like 
it didn't work but I could see that it in fact worked correctly. I used 
the Ada Unchecked conversion.

Mr Carter the Buffer_Pointer is an access  declaration which was limited 
private. And I just had presented it for the possibility that the 
problem should be created by its limited private pro

Re: Put the access value

[briot.emmanuel]

The approach I tend to use is using `System.Address_Image`:


   El  : Buffer_Pointer := LastBuffer; 
   ...

   if El /= null then
      Ada.Text_IO.Put_Line (System.Address_Image (El.all'Address));
   end if;

or if this is for slightly longer term

   function Convert is new Ada.Unchecked_Conversion
     (Buffer_Pointer, System.Address);
   Ada.Text_IO.Put_Line (System.Address_Image (Convert (El));

This is really just for quick debugging, and the code is never (really, I swear) committed... Otherwise, I would go to the trouble of creating an `Image` function and use that

Re: Put the access value

[AdaMagica]

Am Mittwoch, 15. April 2020 09:20:33 UTC+2 schrieb briot....@gmail.com:
> The approach I tend to use is using `System.Address_Image`:

This is not in the RM. It must be GNAT specific, which you should have said.

Re: Put the access value

[J-P. Rosen]

Le 15/04/2020 à 10:02, AdaMagica a écrit :
> Am Mittwoch, 15. April 2020 09:20:33 UTC+2 schrieb briot....@gmail.com:
>> The approach I tend to use is using `System.Address_Image`:
> 
> This is not in the RM. It must be GNAT specific, which you should have said.
> 
True. But for quick debugging, you don't care about portability...

-- 
J-P. Rosen
Adalog
2 rue du Docteur Lombard, 92441 Issy-les-Moulineaux CEDEX
Tel: +33 1 45 29 21 52, Fax: +33 1 45 29 25 00
http://www.adalog.fr

Re: Put the access value

[AdaMagica]

Am Mittwoch, 15. April 2020 10:51:28 UTC+2 schrieb J-P. Rosen:
> Le 15/04/2020 à 10:02, AdaMagica a écrit :
> > Am Mittwoch, 15. April 2020 09:20:33 UTC+2 schrieb briot....@gmail.com:
> >> The approach I tend to use is using `System.Address_Image`:
> > 
> > This is not in the RM. It must be GNAT specific, which you should have said.
> > 
> True. But for quick debugging, you don't care about portability...

True, but there might be some people here, a remote possibility, of utmost improbability, not using GNAT.

Re: Put the access value

[Björn Lundin]

Den 2020-04-15 kl. 10:02, skrev AdaMagica:
> Am Mittwoch, 15. April 2020 09:20:33 UTC+2 schrieb briot....@gmail.com:
>> The approach I tend to use is using `System.Address_Image`:
> 
> This is not in the RM. It must be GNAT specific, which you should have said.
> 

Are you really saying that everything that is posted here
must be checked with the RM?

-- 
Björn

Re: Put the access value

[Randy Brukardt]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 866 bytes --]

"Björn Lundin" <b.f.lundin@gmail.com> wrote in message 
news:r7cdvt$fp5$1@dont-email.me...
> Den 2020-04-15 kl. 10:02, skrev AdaMagica:
>> Am Mittwoch, 15. April 2020 09:20:33 UTC+2 schrieb briot....@gmail.com:
>>> The approach I tend to use is using `System.Address_Image`:
>>
>> This is not in the RM. It must be GNAT specific, which you should have 
>> said.
>>
>
> Are you really saying that everything that is posted here
> must be checked with the RM?

Ideally, everything posted here would note if it is GNAT-specific or 
something defined by Ada, because some of us do primarily use other Ada 
compilers. GNAT /= Ada!

I realize that it isn't always easy to do, so I personally don't complain 
often, but it is definitely helpful to emphasize what is available to all 
Ada users and what is only available to a select group.

                        Randy.

Re: Put the access value

[Björn Lundin]

Den 2020-04-21 kl. 23:06, skrev Randy Brukardt:
> "Björn Lundin" <b.f.lundin@gmail.com> wrote in message
> news:r7cdvt$fp5$1@dont-email.me...
>> Den 2020-04-15 kl. 10:02, skrev AdaMagica:
>>> Am Mittwoch, 15. April 2020 09:20:33 UTC+2 schrieb briot....@gmail.com:
>>>> The approach I tend to use is using `System.Address_Image`:
>>>
>>> This is not in the RM. It must be GNAT specific, which you should have
>>> said.
>>>
>>
>> Are you really saying that everything that is posted here
>> must be checked with the RM?
> 
> Ideally, everything posted here would note if it is GNAT-specific or
> something defined by Ada, because some of us do primarily use other Ada
> compilers. GNAT /= Ada!

I do agree, but the tone of AdaMagica is more like a professor telling a 
student what to do or not.

As if not helping the OP at all is much better than
suggest something not portable, but might be a solution to the 
particular problem, where it is not stated if portability is a 
requirement or not.

Irritating.

-- 
Björn

Re: Put the access value

[AdaMagica]

Am Mittwoch, 22. April 2020 14:31:11 UTC+2 schrieb björn lundin:
> I do agree, but the tone of AdaMagica is more like a professor telling a 
> student what to do or not.

I appoligize if my answer made this impression - this was not my intent.

My thinking was: Someone not using GNAT might think "Oh, that's nice, I'll remember" and later "WTF? Isn't Ada always praised for its portability?"
And there is that habit of many equalling GNAT and Ada (for instance claiming that body files must have the extension adb - even with GNAT, that isn't the case).

Re: Put the access value

[Björn Lundin]

Den 2020-04-22 kl. 19:33, skrev AdaMagica:
> Am Mittwoch, 22. April 2020 14:31:11 UTC+2 schrieb björn lundin:
>> I do agree, but the tone of AdaMagica is more like a professor telling a
>> student what to do or not.
> 
> I appoligize if my answer made this impression - this was not my intent.

Through the written media it is sometimes hard to express intonations 
and it is sometimes hard to interpret the underlying tone of a message.
I could have been more forgiving in my reply,
and it is nice to know that the tone i read into it was not there.


> My thinking was: Someone not using GNAT might think "Oh, that's nice, I'll remember" and later "WTF? Isn't Ada always praised for its portability?"

Well that is true only so far.

While i do agree with Randy that is is good to state the compiler,
I don't think failure to do so should be rewarded with a
'you should have done that'

And is it really enough to say it's gnat?


I don't think the version of the language was mentioned
  (83/95/05/12/15/2x)
(does both 15 and 12 exist?)
I have had case where my collegues used 'Img (gnat-specifik) and
then changed it to 'Image - part of a newer standard - for portability.

Then that code has been moved to another project, where it fails to 
compile. That used gnat 7.4.2, which is 2012. but not 2015. And only
the gnat specific attribute works there.
And here I hear people say 'use type'image(varible) instead.

But I do think

Assignment_Id'image or Assignment_Id'Img is nicer than

with Core_Types
...
Core_Types.Assignment_Identity_Type'Image(Assignment_Id)

which we did before to log out an Assignment_Id


Another thing if working with operating system is the compiler's
runtime tasking model.

A piece of code using blocking calls for I/O in a thread will
work well on implementations mapping tasks to os-threads,
but will block the whole process if the tasking is made by the compiler 
runtime. Alysy Ada did that on AIX (Ada 83). And I think Janus as well.
Object Ada (I think) and Gnat does not.

There are likely other things that break portability

So, should one state language version and runtime behaviour too?


> And there is that habit of many equalling GNAT and Ada (for instance claiming that body files must have the extension adb - even with GNAT, that isn't the case).

Yes, and that is unfortunate.
But critisizing other helpful posts does not change that.
If the OP would have tried the gnat-specific advice given, and failed,
he/she would likely have come back and said 'no good'


-- 
Björn