[Solved] Relationship between high(ashortstring) and length(ashortstring)

[Fred vS]

Hello.

This is a question asked in Spanish forum but we (I) dont get the logical.
https://forum.lazarus.freepascal.org/index.php/topic,60504.msg452955.html#msg452955

With this:

Code: Pascal  [Select][+][-]

1. program cadena;
2. {$mode objfpc}{$H-}  // Alias String is ShortString
3. var
4.   Frase: String;
5. begin
6.   Frase := 'Esto es una prueba';
7.   WriteLn(low(Frase));
8.   WriteLn(High(Frase));
9.   WriteLn(Length(Frase));
10.   Readln()
11. end.
12.  

This is the result:

Code: Bash  [Select][+][-]

1. fred@fred-80m0 ~> ./cadena
2. 0
3. 255
4. 18

Why High(Frase) = 255 (and not Length(Frase) - 1) ?

[KodeZwerg]

Length() used on a string gives back its size in chars.
Low() and High() give back start and end position for an array.

In your case the string is optimized by compiler to a shortstring.

Reading definition for https://wiki.lazarus.freepascal.org/Shortstring shows me:

Warning: low(myShortString) returns 0, i. e. the index of the length Byte, not the first character’s index. Likewise, high(myShortString) always returns 255. Use length instead.

I hope this helps the spanish forum 

[Fred vS]

Warning: ... Likewise, high(myShortString) always returns 255. Use length instead.

Ha, ok, hidden in the bottom, many thanks for the light.
(But, imho, a few exotic to do so, anyway, understood now.)

[Arioch]

confuse them yet better :-)

program cadena;
{$mode objfpc}{$H+}  // Alias String is AnsiString
var
  Frase: String[100];
begin
  Frase := 'Esto es una prueba';
  WriteLn(low(Frase));
  WriteLn(High(Frase));
  WriteLn(Length(Frase));
  Readln()
end.

[Kays]

Low() and High() give back start and end position for an array.

I concur. Any index in the range low(T)..high(T) is valid insofar it will not cause a memory access violation. It doesn’t mean the component at high(T) will be meaningful (has been initialized). To get meaningful content from a shortString use the range 1..length(T) as index.

[Fred vS]

confuse them yet better :-)

program cadena;
{$mode objfpc}{$H+}  // Alias String is AnsiString
var
  Frase: String[100];
begin
  Frase := 'Esto es una prueba';
  WriteLn(low(Frase));
  WriteLn(High(Frase));
  WriteLn(Length(Frase));
  Readln()
end.

And with shortstring:
 

Code: Pascal  [Select][+][-]

1.  program cadena;
2. {$mode objfpc}{$H-} // shortstring
3. var
4.     Frase: String[100];
5. begin
6.   Frase := 'Esto es una prueba';
7.   WriteLn(low(Frase));
8.   WriteLn(High(Frase));
9.   WriteLn(Length(Frase));
10.   Readln()
11. end.
12.  

--->

Code: Bash  [Select][+][-]

1. fred@fred-80m0 ~> ./cadena
2. 0
3. 100
4. 18

Ha, yes, indeed, more confusion. ( in doc: Likewise, high(myShortString) always returns 255.)

[MarkMLl]

Low() and High() give back start and end position for an array.

I concur. Any index in the range low(T)..high(T) is valid insofar it will not cause a memory access violation. It doesn’t mean the component at high(T) will be meaningful (has been initialized). To get meaningful content from a shortString use the range 1..length(T) as index.

To emphasise that: irrespective of what Low() might return, in the case of a string any reference to element [ 0 ] should be treated as an error or- at the very least- extremely ill-advised.

Similarly, and leaving aside encoding problems caused by UTF-8 etc., any reference to an element beyond [ Length() ] should be treated as extremely ill-advised, since it is possible that a string implementation that allocates memory in chunks could return a misleading High().

MarkMLl

[Fred vS]

confuse them yet better :-)

program cadena;
{$mode objfpc}{$H+}  // Alias String is AnsiString

Oops, I did not see that you changed with ansisstring, the problem is mainly with shortstring (ansistring is understood).
Anyway your example gives confusion also.
See my previous post edited.

[Arioch]

There is no confusion for people who remember 16-bit TuboPascal and Delphi 1.x - those did not have LongStrings, and did not needed them very much indeed.

Basically it is the same history that makes LongString (as those types of string are together called in Delphi RTL internals) first letter by s[1] while dynarrays start with a[0]

Actually it was AnsiString introduced that made a lot of confusion then. Suddenyly SizeOf(s) started showing wrong results. Suyddenly s[0] := #13 did not more change string's length, etc.
Frankly, i'd like every programmer read about ShortStrings mandatoriy, because without knowing that history you indeed can not grasp why dynarrays and long strings differ and has to memorize without understanding.

Oh, and more coans.

Code: Pascal  [Select][+][-]

1. var w: WideString; a: AnsiString;
2. begin
3.     a := '1234';  w := a;  // the same value, right ?
4.     WriteLn( SizeOf(a[1]): 10, SizeOf(a): 10);   // expected, right?
5.     WriteLn( SizeOf(w[1]): 10, SizeOf(w): 10);  // did it scale well ?
6. end;
7.  

Ooops, that was for 32-bit code. For you it has to be a := '12345678';  w := a;

When you grasp that history it all is obvious and expected. When you had some experience with that 16-bits programminf - yet better.

[Arioch]

Ha, yes, indeed, more confusion. ( in doc: Likewise, high(myShortString) always returns 255.)

Because by definition type ShortString = string[255];  So simple and no any confusion.

You just have to think not about "magic names" but about internal implementation, data structures. What are  strings of different types in memory, if strings were record or classes - what variable would they ocntain inside?

https://docwiki.embarcadero.com/RADStudio/Alexandria/en/Internal_Data_Formats_(Delphi)
and maybe a bit in https://www.ararat.cz/synapse/doku.php/public:howto:d2009
Modify message

[Fred vS]

Ha, yes, indeed, more confusion. ( in doc: Likewise, high(myShortString) always returns 255.)

Because by definition type ShortString = string[255];  So simple and no any confusion.

The confusion is that in doc, they dont talk about type ShortString but about the variable myShortString that can have, see your demo adapted for shortstring, other high(myshorstring) = 255.
https://forum.lazarus.freepascal.org/index.php/topic,60528.msg452970.html#msg452970

See in bottom of doc:
 https://wiki.lazarus.freepascal.org/Shortstring

[avk]

Just curious, what does the FPC documentation say about the mentioned functions (Low(), High())?

[Arioch]

Just curious, what does the FPC documentation say about the mentioned functions (Low(), High())?

you are welcome - https://docs.getlazarus.org/#rtl+system+high

[avk]

Are you sure this is a link to the official documentation?

[Fred vS]

@Arioch

You seem to be a user of Delphi too.
One of the question in Spanish forum was why is there a difference in result between Delphi for this:

Code: Pascal  [Select][+][-]

1.   program cadena;
2. {$mode objfpc}{$H-}
3. var
4.     Frase: String;
5. begin
6.   Frase := 'Esto es una prueba';
7.   WriteLn(low(Frase));
8.   WriteLn(High(Frase));
9.   WriteLn(Length(Frase));
10.   WriteLn(Frase[500]);
11.   Readln()
12. end.

You may try to compile it on Delphi and compare the result with fpc.
There are also difference in warnings and crash (or not) if range is higher than high(ashortstring).
This is not a problem if we understand the logical of fpc and it is what I try.