Is it possible to define constant records in FPC?

[garlar27]

Is it possible to define constant records in FPC?

If yes: How do you do it?

[Leledumbo]

No, Pascal doesn't have the concept of "constant variables".

[xenblaise]

Is it possible to define constant records in FPC?

Did you mean;
type
pdatas=^datas;
datas=record
id:integer;
name:string;
end;
var mydatas:pdatas;

like that?
Sorry If I got it wrong

[Laksen]

You can

Code: [Select]

type
 TTest = record
  A: string;
  B: longint;
  C: single;
 end;

const
 TestRec: TTest = (A: 'Something'; B: 123; C: 12.3);


[xenblaise]

 

[Leledumbo]

You can

No, that doesn't make the record constant. Borland invented it due to the lack of variable initialization in standard Pascal.

[Laksen]

Sure, it's a typed constant, meaning you can change it

I don't think there's a way to change that

[xenblaise]

type
 TTest = record
  A: string;
  B: longint;
  C: single;
 end;

var xttest:ttest;
...
xttest.a:='somestring';
xttest.b:=12345;
xttest.c:=1;

Did that changed?

It will not changed only if you declare to your write procedure as nil
like;
....no declaration here....//so use the constant
memstream.writebuffer(xttest, sizeof(xttest));

Correct me if I'm wrong...


Thanks

[Leledumbo]

Sure, it's a typed constant, meaning you can change it

I don't think there's a way to change that

Try changing the record attribute, the compiler will allow it. That's "not" constant.

[garlar27]

You can

Code: [Select]

type
 TTest = record
  A: string;
  B: longint;
  C: single;
 end;

const
 TestRec: TTest = (A: 'Something'; B: 123; C: 12.3);


Yeah!! It works!! I didn't figured it out how to do it (at least I couldn't find it in the PDFs), and my guesses didn't compile.

Sure, it's a typed constant, meaning you can change it

I don't think there's a way to change that

Try changing the record attribute, the compiler will allow it. That's "not" constant.

I'm working on a project with lots of constants and I would like to have them grouped (i.e. to have the error message and its error number in the same constant record).

I know it's not a TRUE constant the compiler won't warn you if you change it by code nor at run time, but by changing its capitalization you may know that the var "TEST_REC" is something "special".

But honestly it would be nice to have TRUE constant records.

[Leledumbo]

I know it's not a TRUE constant the compiler won't warn you if you change it by code nor at run time, but by changing its capitalization you may know that the var "TEST_REC" is something "special".

Convention over semantics, that's fine as long as you're consistent.

But honestly it would be nice to have TRUE constant records.

I'd rather name it "read-only variables" since it's more general and is better logically than "constant variables" (a constant shouldn't have variadic values, that's silly). Anyway, I seldom need it and if I do, I'll make it a read-only property of a class.

[captian jaster]

Sure, it's a typed constant, meaning you can change it

I don't think there's a way to change that

This
Though this compiled with no problem:

Code: Pascal  [Select][+][-]

1. program Project1;
2.  
3. {$mode objfpc}{$H+}
4.  
5. uses
6.   Classes, SysUtils;
7.  
8. Const
9.  Trec:Record
10.  S:String;
11.  Int:Integer;
12.  end = (S:'';Int:10);
13.  
14. begin
15. end.
16.  

I know it's not a TRUE constant the compiler won't warn you if you change it by code nor at run time, but by changing its capitalization you may know that the var "TEST_REC" is something "special".

Convention over semantics, that's fine as long as you're consistent.

But honestly it would be nice to have TRUE constant records.

I'd rather name it "read-only variables" since it's more general and is better logically than "constant variables" (a constant shouldn't have variadic values, that's silly). Anyway, I seldom need it and if I do, I'll make it a read-only property of a class.

Buh Buh..
I thought that's what constants were...

[cdbc]

IMHO...
 I would do it the 'Leledumbo' way :-)

Regards Benny

[captian jaster]

IMHO...
 I would do it the 'Leledumbo' way :-)

Regards Benny

Read Only Properties?
Still a variable depending on what your doing...

[garlar27]

   I like the Laksen's approach over Captain Jaster's because you can define many constants of the same record type without repeating its declaration and also you can pass the whole structure as a method's parameter.

   The fact that you can alter that constant value is something that is useful in some cases, but, it's a problem if someone change its value accidentally (since compiler won't warn you).

   I would like to have compile directive (something like: {$+SOMETHING} and {$-SOMETHING}) to allow/prevent modify a typed constant. I don't know if that's possible, nor if that's a good idea ...