Ada.Strings.Unbounded vs Ada.Containers.Indefinite_Holders

Ada.Strings.Unbounded vs Ada.Containers.Indefinite_Holders

[Victor Porton]

I am writing an interface to a C library. Some C functions accept either a C 
string pointer or NULL.

I need some way to define an Ada object which could hold values equivalent 
to C strings with possible NULL.

I consider two approaches:

1. Ada.Strings.Unbounded (remind that it can be Null_Unbounded_String which 
is equivalent to NULL C string).

2. Ada.Containers.Indefinite_Holders holding a String.

The first way seems more natural, as a more specialized package is used.

The second way may have the advantage that there is Is_Empty function while 
for unbounded strings I would need to compare with Null_Unbounded_String 
using "=" operator. This may be a little slower. Or does it matter with 
modern compilers?

Hm well, do I understand Null_Unbounded_String correctly? Is 
Null_Unbounded_String really "no string"? or is it just empty string ("")? 
RM2012 should be made more clear on this topic!

-- 
Victor Porton - http://portonvictor.org

Re: Ada.Strings.Unbounded vs Ada.Containers.Indefinite_Holders

[AdaMagica]

Am Donnerstag, 21. September 2017 20:14:29 UTC+2 schrieb Victor Porton:
> This may be a little slower.

Why do you care for performance? Have you ever measured what you are worrying about?

Premature optimization is the source of all evil - you can hear this mantra very often here.


> Hm well, do I understand Null_Unbounded_String correctly? Is 
> Null_Unbounded_String really "no string"? or is it just empty string ("")? 

It's the latter. Is RM A.4.5(73) and 2.5(6) clear enough?

Re: Ada.Strings.Unbounded vs Ada.Containers.Indefinite_Holders

[Victor Porton]

AdaMagica wrote:

> Am Donnerstag, 21. September 2017 20:14:29 UTC+2 schrieb Victor Porton:
>> This may be a little slower.
> 
> Why do you care for performance? Have you ever measured what you are
> worrying about?
> 
> Premature optimization is the source of all evil - you can hear this
> mantra very often here.
> 
> 
>> Hm well, do I understand Null_Unbounded_String correctly? Is
>> Null_Unbounded_String really "no string"? or is it just empty string
>> ("")?
> 
> It's the latter. Is RM A.4.5(73) and 2.5(6) clear enough?

The problem I asked about is exactly that A.4.5(73) is not clear enough. 
What is 2.5(6)? There is no (6) in RM 2.5.

Well, you are right (assuming that GCC 7.2.0 does the right thing), as this 
prints "True":

with Ada.Strings.Unbounded; use Ada.Strings.Unbounded;
with Ada.Text_IO;

procedure Test is
begin
  Ada.Text_IO.Put_Line(Boolean'Image(To_Unbounded_String("") = 
Null_Unbounded_String));
end;


-- 
Victor Porton - http://portonvictor.org

Re: Ada.Strings.Unbounded vs Ada.Containers.Indefinite_Holders

[Simon Wright]

Victor Porton <porton@narod.ru> writes:

> What is 2.5(6)? There is no (6) in RM 2.5.

2.6(6):

    A null string literal is a string_literal with no string_elements
    between the quotation marks.

Re: Ada.Strings.Unbounded vs Ada.Containers.Indefinite_Holders

[Victor Porton]

Simon Wright wrote:

> Victor Porton <porton@narod.ru> writes:
> 
>> What is 2.5(6)? There is no (6) in RM 2.5.
> 
> 2.6(6):
> 
>     A null string literal is a string_literal with no string_elements
>     between the quotation marks.

In my opinion, it would be better to change RM phrasing from "null string" 
to "empty string", because in some other languages (notably C) NULL means 
something other. It is just confusing.

-- 
Victor Porton - http://portonvictor.org

Re: Ada.Strings.Unbounded vs Ada.Containers.Indefinite_Holders

[Dmitry A. Kazakov]

On 2017-09-23 00:15, Victor Porton wrote:
> Simon Wright wrote:
> 
>> Victor Porton <porton@narod.ru> writes:
>>
>>> What is 2.5(6)? There is no (6) in RM 2.5.
>>
>> 2.6(6):
>>
>>      A null string literal is a string_literal with no string_elements
>>      between the quotation marks.
> 
> In my opinion, it would be better to change RM phrasing from "null string"
> to "empty string", because in some other languages (notably C) NULL means
> something other. It is just confusing.

The adjective null and the noun null are distinct parts of speech. C's 
noun null is an abbreviation of null pointer. If pointers can be null so 
strings can.

Wouldn't it be silly to change, say, mathematical group theory with its 
null group elements because C (and Ada too) have null pointers called nulls?

BTW, empty is string is indeed a null element for the concatenation 
operation &:

    forall X in String   X & "" = X

So, using the null adjective is absolutely justified.

A null string is a bit more than just empty string. (:-))

-- 
Regards,
Dmitry A. Kazakov
http://www.dmitry-kazakov.de

Re: Ada.Strings.Unbounded vs Ada.Containers.Indefinite_Holders

[Jeffrey R. Carter]

On 09/23/2017 10:09 AM, Dmitry A. Kazakov wrote:
> On 2017-09-23 00:15, Victor Porton wrote:
>>
>> In my opinion, it would be better to change RM phrasing from "null string"
>> to "empty string", because in some other languages (notably C) NULL means
>> something other. It is just confusing.
> 
> The adjective null and the noun null are distinct parts of speech. C's noun null 
> is an abbreviation of null pointer. If pointers can be null so strings can.

Another way to look at it: Ada has the formal concepts of:

* null access value ARM 4.2(9)
* null array 3.6.1(7)
* null constraint 3.2(7/2)
* null_exclusion 3.10(5.1/2)
* null extension 3.9.1(4.1/2)
* null procedure 6.7(3/3)
* null range 3.5(4)
* null record 3.8(15)
* null slice 4.1.2(7)
* null string literal 2.6(6)
* null value (of an access type) 3.10(13/2)
* null_statement 5.1(6)

not to mention the language-defined identifiers

Null_Address
    in System   13.7(12)
Null_Bounded_String
    in Ada.Strings.Bounded   A.4.4(7)
Null_Id
    in Ada.Exceptions   11.4.1(2/2)
Null_Occurrence
    in Ada.Exceptions   11.4.1(3/2)
Null_Ptr
    in Interfaces.C.Strings   B.3.1(7)
Null_Set
    in Ada.Strings.Maps   A.4.2(5)
    in Ada.Strings.Wide_Maps   A.4.7(5)
    in Ada.Strings.Wide_Wide_Maps   A.4.8(5/2)
Null_Task_Id
    in Ada.Task_Identification   C.7.1(2/2)
Null_Unbounded_String
    in Ada.Strings.Unbounded   A.4.5(5)

(Just look under N in the index.)

It's called overloading. Many of these cases refer to things that can have 
components and mean one with zero components: a null record has no components, a 
null array has no components ('Length = 0), a null string literal has no 
characters, a null set has no members, ... It should not be confusing.

-- 
Jeff Carter
"You cheesy lot of second-hand electric donkey-bottom biters."
Monty Python & the Holy Grail
14

Re: Ada.Strings.Unbounded vs Ada.Containers.Indefinite_Holders

[Randy Brukardt]

"Jeffrey R. Carter" <spam.jrcarter.not@spam.not.acm.org> wrote in message 
news:oq58qb$o3s$1@dont-email.me...
> On 09/23/2017 10:09 AM, Dmitry A. Kazakov wrote:
>> On 2017-09-23 00:15, Victor Porton wrote:
>>>
>>> In my opinion, it would be better to change RM phrasing from "null 
>>> string"
>>> to "empty string", because in some other languages (notably C) NULL 
>>> means
>>> something other. It is just confusing.
....
> It's called overloading. Many of these cases refer to things that can have 
> components and mean one with zero components: a null record has no 
> components, a null array has no components ('Length = 0), a null string 
> literal has no characters, a null set has no members, ... It should not be 
> confusing.

And it also shows the amount of change needed to actually make such a change 
to the RM. It would be a lot of change for (at most) not much value, so it 
wouldn't be a very good use of scarce resources.

                              Randy.