Call for testing: array constructors

classic Classic list List threaded Threaded
9 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Call for testing: array constructors

Free Pascal - General mailing list
Hello together!

Since revision 36105 FPC now supports the use of array constructors
using the "[...]" syntax inside ordinary code blocks like Delphi does
since - I think - XE8. And yes, even nested ones are supported (take a
look at $fpcdir/tests/test/tarrconstr5.pp for a bit of inspiration).

Considering that this changed how "[...]" is handled I'd like you all to
test whether your existing code still works (especially if it's dealing
with sets!) and to try this new feature to see if there are any problems
that our testsuite doesn't cover yet.

If you report bugs, then please attach the tag "array constructors".

Have fun! :D

Regards,
Sven
_______________________________________________
fpc-pascal maillist  -  [hidden email]
http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Call for testing: array constructors

Mark Morgan Lloyd-5
On 04/05/17 22:30, Sven Barth via fpc-pascal wrote:
> Hello together!
> Since revision 36105 FPC now supports the use of array constructorsusing the "[...]" syntax inside ordinary code blocks like Delphi doessince - I think - XE8. And yes, even nested ones are supported (take alook at $fpcdir/tests/test/tarrconstr5.pp for a bit of inspiration).
> Considering that this changed how "[...]" is handled I'd like you all totest whether your existing code still works (especially if it's dealingwith sets!) and to try this new feature to see if there are any problemsthat our testsuite doesn't cover yet.
> If you report bugs, then please attach the tag "array constructors".

Ah yes, /very/ nice :-)

I append a chunk of fun code, which as it stands needs separate
functions per rank (i.e. for 1 dimension, 2 dimensions and so on). Can
these be rationalised using generics?

Output should look something like

a3:
1 2 3 4 5
1 2 3 4 5
1 2 3 4 5

5 6 7 8 9
1 3 5 7 9
2 4 6 8 0


+/ a3:
6 8 10 12 14
2 5 8 11 14
3 6 9 12 5


a2:
6 8 10 12 14
2 5 8 11 14
3 6 9 12 5

+/ a2:
11 19 27 35 33


a1:
11 19 27 35 33
+/ a1:
125

--
Mark Morgan Lloyd
markMLl .AT. telemetry.co .DOT. uk

[Opinions above are the author's, not those of his employers or colleagues]




program testReduction;

{$mode objfpc}{$H+}

// This demo program abuses FPC's custom operator facility to simulate
// Vector Pascal's \+ (reduce-add) operator, which is derived from APL's
// +/ function/operator.

// It's reasonable to expect no sane code to try to attempt to add a
// procedure to an array or a number.

type     TReduce= procedure(p1: pointer);

const    Reduce: TReduce= nil;

type    t1= array of longint;
         t2= array of t1;
         t3= array of t2;

{$MACRO ON}
{$DEFINE SHOW_IDENTITY:=// WriteLn('Identity:'); print(result); WriteLn; }


procedure print(i: longint);

begin
   Write(i)
end { print } ;


procedure print(const a: t1);

var     i: integer;

begin
   for i := Low(a) to High(a) do
     Write(a[i], ' ');
   WriteLn
end { print } ;


procedure print(const a: t2);

var     i: integer;

begin
   for i := Low(a) to High(a) do
     print(a[i]);
   WriteLn
end { print } ;


procedure print(const a: t3);

var     i: integer;

begin
   for i := Low(a) to High(a) do
     print(a[i]);
//  WriteLn('-----');
   WriteLn
end { print } ;


operator + (const a1, a2: t1): t1;

var     i: integer;

begin
   SetLength(result, Length(a1));
   for i := Low(a1) to High(a1) do
     result[i] := a1[i] + a2[i]
end { + } ;


operator + (const a1, a2: t2): t2;

var     i: integer;

begin
   SetLength(result, Length(a1));
   for i := Low(a1) to High(a1) do
     result[i] := a1[i] + a2[i]
end { + } ;


const    additiveIdentity= 0;
          multiplicativeIdentity= 1;


operator + (const r: TReduce; const a: t1): longint;

var     i: integer;

begin
   result := additiveIdentity;
   SHOW_IDENTITY
   for i := Low(a) to High(a) do
     result += a[i]
end { + } ;


operator + (const r: TReduce; const a: t2): t1;

var     i: integer;

begin
   SetLength(result, Length(a[Low(a)]));
   for i := Low(result) to High(result) do
     result[i] := additiveIdentity;
   SHOW_IDENTITY
   for i := Low(a) to High(a) do
     result += a[i]
end { + } ;


operator + (const r: TReduce; const a: t3): t2;

var     i: integer;


   function additiveIdentityArray1: t1;

   var   i: integer;

   begin
     SetLength(result, Length(a[0, 0]));
     for i := Low(result) to High(result) do
       result[i] := additiveIdentity
   end { additiveIdentityArray1 } ;


begin
   SetLength(result, Length(a[Low(a)]));
   for i := Low(result) to High(result) do
     result[i] := additiveIdentityArray1;
   SHOW_IDENTITY
   for i := Low(a) to High(a) do begin
     result += a[i]
   end
end { + } ;


var      a0: longint;
          a1: t1;
          a2: t2;
          a3: t3;

begin
   a3 := [[[1,2,3,4,5],[1,2,3,4,5],[1,2,3,4,5]],
          [[5,6,7,8,9],[1,3,5,7,9],[2,4,6,8,0]]];
   WriteLn('a3:');
   print(a3);
   WriteLn('+/ a3:');
   a2 := reduce + a3;
   print(a2); WriteLn;

   WriteLn('a2:');
   print(a2);
   WriteLn('+/ a2:');
   a1 := reduce + a2;
   print(a1); WriteLn;
   WriteLn;

   WriteLn('a1:');
   print(a1);
   WriteLn('+/ a1:');
   a0 := reduce + a1;
   WriteLn(a0);
   WriteLn
end.




_______________________________________________
fpc-pascal maillist  -  [hidden email]
http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Call for testing: array constructors

Free Pascal - General mailing list

Am 05.05.2017 21:19 schrieb "Mark Morgan Lloyd" <[hidden email]>:
>
> On 04/05/17 22:30, Sven Barth via fpc-pascal wrote:
>>
>> Hello together!
>> Since revision 36105 FPC now supports the use of array constructorsusing the "[...]" syntax inside ordinary code blocks like Delphi doessince - I think - XE8. And yes, even nested ones are supported (take alook at $fpcdir/tests/test/tarrconstr5.pp for a bit of inspiration).
>> Considering that this changed how "[...]" is handled I'd like you all totest whether your existing code still works (especially if it's dealingwith sets!) and to try this new feature to see if there are any problemsthat our testsuite doesn't cover yet.
>>
>> If you report bugs, then please attach the tag "array constructors".
>
>
> Ah yes, /very/ nice :-)
>
> I append a chunk of fun code, which as it stands needs separate functions per rank (i.e. for 1 dimension, 2 dimensions and so on). Can these be rationalised using generics?

Generics are only useful as long as you can use the same code for different types. E.g. if you replace the Writeln inside the loop for t1's print with Print(a[i]) then you could at least generalize the print functions.
Operators however would need you to stuff them into a record as only then you could define generic operators that would work on that record type. Alse the code of your addition operators differs with the types so that would not help, at least not as is.

Regards,
Sven


_______________________________________________
fpc-pascal maillist  -  [hidden email]
http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Call for testing: array constructors

Mark Morgan Lloyd-5
In reply to this post by Free Pascal - General mailing list
On 04/05/17 22:30, Sven Barth via fpc-pascal wrote:
> Hello together!
> Since revision 36105 FPC now supports the use of array constructorsusing the "[...]" syntax inside ordinary code blocks like Delphi doessince - I think - XE8. And yes, even nested ones are supported (take alook at $fpcdir/tests/test/tarrconstr5.pp for a bit of inspiration).
> Considering that this changed how "[...]" is handled I'd like you all totest whether your existing code still works (especially if it's dealingwith sets!) and to try this new feature to see if there are any problemsthat our testsuite doesn't cover yet.
> If you report bugs, then please attach the tag "array constructors".

 > Operators however would need you to stuff them into a record as only
 > then you could define generic operators that would work on that
 > record type. Alse the code of your addition operators differs with
 > the types so that would not help, at least not as is.

Thanks Sven. Am I correct in believing that operators are basically not
handled by generics?

Apart from that, the operations of getting stuff into arrays etc. appear
to work well.

--
Mark Morgan Lloyd
markMLl .AT. telemetry.co .DOT. uk

[Opinions above are the author's, not those of his employers or colleagues]
_______________________________________________
fpc-pascal maillist  -  [hidden email]
http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Call for testing: array constructors

Free Pascal - General mailing list

Am 08.05.2017 16:34 schrieb "Mark Morgan Lloyd" <[hidden email]>:
>
> On 04/05/17 22:30, Sven Barth via fpc-pascal wrote:
>>
>> Hello together!
>> Since revision 36105 FPC now supports the use of array constructorsusing the "[...]" syntax inside ordinary code blocks like Delphi doessince - I think - XE8. And yes, even nested ones are supported (take alook at $fpcdir/tests/test/tarrconstr5.pp for a bit of inspiration).
>> Considering that this changed how "[...]" is handled I'd like you all totest whether your existing code still works (especially if it's dealingwith sets!) and to try this new feature to see if there are any problemsthat our testsuite doesn't cover yet.
>>
>> If you report bugs, then please attach the tag "array constructors".
>
>
> > Operators however would need you to stuff them into a record as only
> > then you could define generic operators that would work on that
> > record type. Alse the code of your addition operators differs with
> > the types so that would not help, at least not as is.
>
> Thanks Sven. Am I correct in believing that operators are basically not handled by generics?

They are handled in so far that operator overloads in records are respected as well if the operator was available at the time the generic was *declared* (cause that scope will be used at specialization time). I have plans to test a change in this, but I don't know yet whether it should make it into trunk...

> Apart from that, the operations of getting stuff into arrays etc. appear to work well.

Good to know. Any regressions with ordinary sets?

Regards,
Sven


_______________________________________________
fpc-pascal maillist  -  [hidden email]
http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Call for testing: array constructors

Mattias Gaertner
In reply to this post by Free Pascal - General mailing list
On Fri, 5 May 2017 00:06:25 +0200
Sven Barth via fpc-pascal <[hidden email]> wrote:

>[...]
> Since revision 36105 FPC now supports the use of array constructors
> using the "[...]" syntax inside ordinary code blocks like Delphi does
> since - I think - XE8. And yes, even nested ones are supported (take a
> look at $fpcdir/tests/test/tarrconstr5.pp for a bit of inspiration).

Do they work the same in mode objfpc?

Mattias

_______________________________________________
fpc-pascal maillist  -  [hidden email]
http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Call for testing: array constructors

Free Pascal - General mailing list

Am 08.05.2017 18:25 schrieb "Mattias Gaertner" <[hidden email]>:
>
> On Fri, 5 May 2017 00:06:25 +0200
> Sven Barth via fpc-pascal <[hidden email]> wrote:
>
> >[...]
> > Since revision 36105 FPC now supports the use of array constructors
> > using the "[...]" syntax inside ordinary code blocks like Delphi does
> > since - I think - XE8. And yes, even nested ones are supported (take a
> > look at $fpcdir/tests/test/tarrconstr5.pp for a bit of inspiration).
>
> Do they work the same in mode objfpc?

Yes, there isn't any mode checking in place for them.

Regards,
Sven


_______________________________________________
fpc-pascal maillist  -  [hidden email]
http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Call for testing: array constructors

Jonas Maebe-3
In reply to this post by Free Pascal - General mailing list
On 08/05/17 17:34, Sven Barth via fpc-pascal wrote:
> Good to know. Any regressions with ordinary sets?

There seems to be a problem with overload selection between dynamic and
open arrays for empty array parameters:
https://bugs.freepascal.org/view.php?id=31756


Jonas
_______________________________________________
fpc-pascal maillist  -  [hidden email]
http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Call for testing: array constructors

Free Pascal - General mailing list

Am 08.05.2017 22:14 schrieb "Jonas Maebe" <[hidden email]>:
>
> On 08/05/17 17:34, Sven Barth via fpc-pascal wrote:
>>
>> Good to know. Any regressions with ordinary sets?
>
>
> There seems to be a problem with overload selection between dynamic and open arrays for empty array parameters: https://bugs.freepascal.org/view.php?id=31756
>

Why am I not surprised? -.- Thanks.

Regards,
Sven


_______________________________________________
fpc-pascal maillist  -  [hidden email]
http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Loading...