Re: [fpc-devel] tbits.NotBits

2017-04-13 Thread Mattias Gaertner 
On Thu, 13 Apr 2017 11:28:20 +0200
Andrea Mauri  wrote:

>[...]
> procedure TBits.Setall;
> var
> loop : longint;
> begin
> for loop := 0 to FSize - 1 do
>FBits^[loop] := 1;

Should be
FBits^[loop] := not cardinal(0);

> end;

Btw, it seems TBits misses an AssignBits(b: TBits).

And maybe an AssignNotBits(b: TBits)


> 2. a Not operator, something like a.NotBits; with no arguments that 
> perform the not operation on the TBIts instance (a[i]:= not a[i]) or 
> something that performs a not operation on the passed TBits, i.e. 
> a.Not(b) that fills a in this way: a[i]:= not b[i];
> 
> procedure TBits.Notbits;
> var
> n : longint;
> jj : cardinal;
> loop : longint;
> begin
> for loop := 0 to FSize - 1 do
>FBits^[loop] := not FBits^[loop];
> end;
> 
> or
> 
> procedure TBits.Not(bitset : TBits);

"Not" is a preserved word.


> var
> n : longint;
> loop : longint;
> begin
> if FSize < bitset.getFSize then
>n := FSize - 1
> else
>n := bitset.getFSize - 1;
> 
> for loop := 0 to n do
>FBits^[loop] := not bitset.FBits^[loop];
> end;

For consistency with the other SomethingBits methods you must handle all
bits, not just the start.


Mattias
___
fpc-devel maillist  -  fpc-devel@lists.freepascal.org
http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-devel

Re: [fpc-devel] tbits.NotBits

2017-04-13 Thread Andrea Mauri 

AFAIK a.NotBits(b) means (a and not b):

a b result
0 0 0
0 1 0
1 0 1
1 1 0

Mattias


Thank you Mattias.
It works as you described.
Anyway, since ClearAll, AndBits, OrBits etc works iterating on
FBits : ^TBitArray;
they are much faster then the simple iteration along property
property Bits[Bit: longint]: Boolean read get write SetBit; default;

I think should be useful to have also at least two other methods 
directly implemented in tbits:
1. the opposite of ClearAll, something like SetAll that sets all bits to 
1, I think this s a typical operation on Bits;



procedure TBits.Setall;
var
   loop : longint;
begin
   for loop := 0 to FSize - 1 do
  FBits^[loop] := 1;
end;


2. a Not operator, something like a.NotBits; with no arguments that 
perform the not operation on the TBIts instance (a[i]:= not a[i]) or 
something that performs a not operation on the passed TBits, i.e. 
a.Not(b) that fills a in this way: a[i]:= not b[i];


procedure TBits.Notbits;
var
   n : longint;
   jj : cardinal;
   loop : longint;
begin
   for loop := 0 to FSize - 1 do
  FBits^[loop] := not FBits^[loop];
end;

or

procedure TBits.Not(bitset : TBits);
var
   n : longint;
   loop : longint;
begin
   if FSize < bitset.getFSize then
  n := FSize - 1
   else
  n := bitset.getFSize - 1;

   for loop := 0 to n do
  FBits^[loop] := not bitset.FBits^[loop];
end;

___
fpc-devel maillist  -  fpc-devel@lists.freepascal.org
http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-devel

Re: [fpc-devel] tbits.NotBits

2017-04-13 Thread Mattias Gaertner 
On Thu, 13 Apr 2017 10:52:44 +0200 (CEST)
Michael Van Canneyt  wrote:

>[...]
> I am not sure that what you did is supported.
> 
> b.notbits(b)
> 
> I am not sure that you can pass the same instance to b.

Just look at the code. It's only a few lines. It is supported.

Mattias
___
fpc-devel maillist  -  fpc-devel@lists.freepascal.org
http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-devel

Re: [fpc-devel] tbits.NotBits

2017-04-13 Thread Michael Van Canneyt 



On Thu, 13 Apr 2017, Andrea Mauri wrote:


any answer?
I asked in the wrong place?
where should I ask?


You asked in the right place. I just didn't notice your first mail.

I am not sure that what you did is supported.

b.notbits(b)

I am not sure that you can pass the same instance to b.

Can you test with 2 separate instances ?

Michael.


Il 31/03/2017 14:10, Andrea Mauri ha scritto:

one more thing.
there is a method like clearall to set all bits to 1?
clearall is much more faster then a simple iteration along all bits

Il 31/03/2017 14:07, Andrea Mauri ha scritto:

I didn't understand how TBits.NotBits works.
Consider the following code:


  b:= TBits.Create(NRECORDS);
  b.NotBits(b);

or more clear:

  b:= TBits.Create(NRECORDS);
  b.Clearall;
  b.NotBits(b);

I supposed that all bits of b will be 1. What I obtain is that all bits
are 0.

fpc 3.0.0 win64

Am I missing something?
Thank you.
Andrea Mauri

___
fpc-devel maillist  -  fpc-devel@lists.freepascal.org
http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-devel


___
fpc-devel maillist  -  fpc-devel@lists.freepascal.org
http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-devel

Re: [fpc-devel] tbits.NotBits

2017-04-13 Thread Mattias Gaertner 
On Fri, 31 Mar 2017 14:07:09 +0200
Andrea Mauri  wrote:

> I didn't understand how TBits.NotBits works.

AFAIK a.NotBits(b) means (a and not b):

a b result
0 0 0
0 1 0
1 0 1
1 1 0

Mattias
___
fpc-devel maillist  -  fpc-devel@lists.freepascal.org
http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-devel

Re: [fpc-devel] tbits.NotBits

2017-04-13 Thread Andrea Mauri 

any answer?
I asked in the wrong place?
where should I ask?

Il 31/03/2017 14:10, Andrea Mauri ha scritto:

one more thing.
there is a method like clearall to set all bits to 1?
clearall is much more faster then a simple iteration along all bits

Il 31/03/2017 14:07, Andrea Mauri ha scritto:

I didn't understand how TBits.NotBits works.
Consider the following code:


  b:= TBits.Create(NRECORDS);
  b.NotBits(b);

or more clear:

  b:= TBits.Create(NRECORDS);
  b.Clearall;
  b.NotBits(b);

I supposed that all bits of b will be 1. What I obtain is that all bits
are 0.

fpc 3.0.0 win64

Am I missing something?
Thank you.
Andrea Mauri

___
fpc-devel maillist  -  fpc-devel@lists.freepascal.org
http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-devel

Re: [fpc-devel] tbits.NotBits

2017-03-31 Thread Andrea Mauri 
one last thing, using TBits it is possible to SetIndex in order to start 
from a predefined index when using FindNextBit and FindPrevBit
Anyway it is possible to set the State only using FindFirstBit, so I 
cannot use FindNextBit only using SetIndex but I have to call also 
FindFirstBit.

Is it possible to introduce a SetState procedure?
Thanks again,
Andrea Mauri

Il 31/03/2017 14:10, Andrea Mauri ha scritto:

one more thing.
there is a method like clearall to set all bits to 1?
clearall is much more faster then a simple iteration along all bits

Il 31/03/2017 14:07, Andrea Mauri ha scritto:

I didn't understand how TBits.NotBits works.
Consider the following code:


  b:= TBits.Create(NRECORDS);
  b.NotBits(b);

or more clear:

  b:= TBits.Create(NRECORDS);
  b.Clearall;
  b.NotBits(b);

I supposed that all bits of b will be 1. What I obtain is that all bits
are 0.

fpc 3.0.0 win64

Am I missing something?
Thank you.
Andrea Mauri

___
fpc-devel maillist  -  fpc-devel@lists.freepascal.org
http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-devel

Re: [fpc-devel] tbits.NotBits

2017-03-31 Thread Andrea Mauri 

one more thing.
there is a method like clearall to set all bits to 1?
clearall is much more faster then a simple iteration along all bits

Il 31/03/2017 14:07, Andrea Mauri ha scritto:

I didn't understand how TBits.NotBits works.
Consider the following code:


  b:= TBits.Create(NRECORDS);
  b.NotBits(b);

or more clear:

  b:= TBits.Create(NRECORDS);
  b.Clearall;
  b.NotBits(b);

I supposed that all bits of b will be 1. What I obtain is that all bits
are 0.

fpc 3.0.0 win64

Am I missing something?
Thank you.
Andrea Mauri

___
fpc-devel maillist  -  fpc-devel@lists.freepascal.org
http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-devel