IfThenStr does not work - omg

[jamie]

which raises the question from my end, since I've been out of commission for some months limiting my time here.

Has the compiler introduced a real Iften function ?, you know one that only evaluates the results of the answer not before hand>

for example:

Code: Pascal  [Select][+][-]

1. SomeNeededResults := If theCondition Then TrueAnswerFunction else falseAnswerFunciton;
2.  

Jamie

[Zoran]

which raises the question from my end, since I've been out of commission for some months limiting my time here.

Has the compiler introduced a real Iften function ?, you know one that only evaluates the results of the answer not before hand>

for example:

Code: Pascal  [Select][+][-]

1. SomeNeededResults := If theCondition Then TrueAnswerFunction else falseAnswerFunciton;
2.  

Jamie

Unfortunately not. PascalDragon implemented it long time ago as an intrinsic function, but it was rejected by other developers.
See: https://lists.freepascal.org/pipermail/fpc-pascal/2016-January/046376.html

[jamie]

IF that is the case, then I believe FP needs to put it back in.

There is already ifThen in the libs, but they evaluate the resulting functions before the condition is known, since it is a function.

Proper handling of this code would allow only the function needed to get executed which solves two problems, unwanted call to the other function
and execution speed.

 Using the Case state does this but it does not return a results.

 this of course, like you say would be an intrinsic operation.

 I invision the days where I can use this code within prototype calls as functions are called on the fly..

Code: Pascal  [Select][+][-]

1.  
2.  SomeFunction (If Condition then TrueFunction else FalseFunction);
3.  
4.  
5.  

 oh well.

Jamie

[paule32]

you coming from C ?

[jamie]

you coming from C ?

What gave it away?

jamie

[Paolo]

Unfortunately not. PascalDragon implemented it long time ago as an intrinsic function, but it was rejected by other developers

Maybe it is time to renew the feature request?

[CM630]

 

Code: Pascal  [Select][+][-]

1. ShowMessage(ifthen(true,'true','false'));

results in
 

Error: Generics without specialization cannot be used as a type for a variable

[Zoran]

 

Code: Pascal  [Select][+][-]

1. ShowMessage(ifthen(true,'true','false'));

results in
 

Error: Generics without specialization cannot be used as a type for a variable

There will be Implicit specialization in next major version of fpc, which might arrive even in our grandchildren's lives.

[ALLIGATOR]

...which might arrive even in our grandchildren's lives.

😅➡️😟

[MarkMLl]

I agree with Jamie.

you coming from C ?

No, ALGOL-60. Wirth omitted it when he developed Pascal, which was (for political reasons) a rush job.

MarkMLl

[Thaddy]

Yes, my patch does indeed work with implicit specialization, but note that the implicit specialization feature is not always stable yet. There are quite some caveats in current main and it is quite hard to debug what's going wrong.
This works:

Code: Pascal  [Select][+][-]

1. {$if fpc_fullversion < 30301}{$error this code needs 3.3.1 or higher}{$ifend}
2. {$mode delphi}{$modeswitch implicitfunctionspecialization}
3. uses sysutils;
4. type
5.   Ttest = function:string;
6.  
7. function SomeTestOne:string;
8. begin
9.  writestr(result,'Test1');
10. end;
11.  
12. function SomeTestTwo:string;
13. begin
14.  writestr(result,'Test2')
15. end;  
16.  
17. var
18.   a:integer = 100;
19.   b:integer = 200;
20.   c:single = 0.001;
21.   d:single = 1.000;
22.   e:TTest = SomeTestOne;
23.   f:TTest = SomeTestTwo;
24. begin
25.   // all these use sysutils.ifthen<T>
26.   // using implicit specialization
27.   writeln(ifthen(true,a,b));
28.   writeln(ifthen(false,a,b));
29.   writeln(ifthen(true,c,d):2:3);
30.   writeln(ifthen(false,c,d):2:3);
31.   writeln(ifthen(true,e,f));  // executes e
32.   writeln(ifthen(false,e,f)); // executes f
33.   writeln(ifthen(true,'true','false'));
34.   writeln(ifthen(false,'true','false'));  
35. end.

But some scenario's can lead to known problems, merely related to scoping and parameter sizes.
E.g. if math and/or strutils are also used.
A solution is to include ifthen<T> in system instead, but in that case PascalDragon's intrinsic would be a much better solution than a generic.

[Thaddy]

Some pitfalls are discussed here:
https://forum.lazarus.freepascal.org/index.php?topic=59124.0
Few have been fixed yet.