We use latest Delphi(10.2 Tokyo and latest FastMM4

Code is following

function BigIntegerStrMod(const ABigInteger: string; const Adivider: Integer): Integer;
var
LBigInt: BigInteger;
LDivisor: BigInteger;
LReminder: BigInteger;
begin
// Alkuarvo
if not BigInteger.TryParse(ABigInteger, LBigInt) then
raise Exception.CreateFmt('Virheellinen numero %s', [ABigInteger]);
LDivisor := BigInteger.Create(Adivider);
LReminder := BigInteger.Create(0);

// Lasketana ja Palautetaan
LReminder := LBigInt mod LDivisor;
Result := LReminder.AsInteger;
end;

At last line when some cleanup happens we get FastMM error

STCommon.UnitTests.exe: Memory Error Detected

FastMM has detected an error during a FreeMem operation. The block footer has been corrupted.

The block size is: 40

This block was allocated by thread 0x54E8, and the stack trace (return addresses) at the time was:

43BDFA [FastMM4][DebugAllocMem$qqri]

409C86 [System.pas][System][AllocMem$qqri][4688]

E2E558 [Velthuis.BigIntegers.pas][Velthuis.BigIntegers][Bigintegers.AllocNewMagnitude$qqrrpvi][9265]

E2E5A5 [Velthuis.BigIntegers.pas][Velthuis.BigIntegers][Bigintegers.BigInteger.MakeSize$qqri][9274]

E2AF40 [Velthuis.BigIntegers.pas][Velthuis.BigIntegers][Bigintegers.BigInteger.TryParse$qqrx20System.UnicodeStringzcr31Velthuis.Bigintegers.BigInteger][5190]

409CBA [System.pas][System][@getmem$qqri][4745]

E2AE55 [Velthuis.BigIntegers.pas][Velthuis.BigIntegers][Bigintegers.BigInteger.TryParse$qqrx20System.UnicodeStringr31Velthuis.Bigintegers.BigInteger][5171]

4105E6 [System.pas][System][@FinalizeArray$qqrpvt1ui][32099]

4104A1 [System.pas][System][@FinalizeRecord$qqrpvt1][31777]

E33CE0 [STCommon.PankkiUtils.pas][STCommon.PankkiUtils][Pankkiutils.BigIntegerStrMod$qqrx20System.UnicodeStringxi][1075]

E33F01 [STCommon.PankkiUtils.pas][STCommon.PankkiUtils][Pankkiutils.TarkistaRFViite$qqrx20System.UnicodeString][1127]

The block is currently used for an object of class: Unknown

The allocation number is: 589545

The block was previously freed by thread 0x54E8, and the stack trace (return addresses) at the time was:

43BD75 [FastMM4][DebugReallocMem$qqrpvi]

409CF7 [System.pas][System][@ReallocMem$qqrrpvi][4867]

40FB19 [System.pas][System][@UStrSetLength$qqrr20System.UnicodeStringi][29850]

40FB9A [System.pas][System][@UStrCat$qqrr20System.UnicodeStringx20System.UnicodeString][29963]

E33C4D [STCommon.PankkiUtils.pas][STCommon.PankkiUtils][Pankkiutils.RFReferenceCharsToNumbers$qqrx20System.UnicodeString][1064]

E33E5C [STCommon.PankkiUtils.pas][STCommon.PankkiUtils][Pankkiutils.NormalizeRFReference$qqrr20System.UnicodeString][1096]

E33EDA [STCommon.PankkiUtils.pas][STCommon.PankkiUtils][Pankkiutils.TarkistaRFViite$qqrx20System.UnicodeString][1120]

4105D5 [System.pas][System][@FinalizeArray$qqrpvt1ui][32070]

E3BD9D [STCommon.PankkiUtils.DUnitX.pas][STCommon.PankkiUtils.DUnitX][Pankkiutils.Dunitx.TSTCommonPankkiUtilsTests.TarkistaRFViiteTest$qqrx20System.UnicodeStringxo][47]

4B6810 [System.Rtti.pas][System.Rtti][Rtti.RawInvoke$qqrpvp23System.Rtti.TParamBlock][7750]

4B6C14 [System.Rtti.pas][System.Rtti][Rtti.Invoke$qqrpvx42System.%DynamicArray$18System.Rtti.TValue%24System.Typinfo.TCallConvp24System.Typinfo.TTypeInfooo][7972]

The current thread ID is 0x54E8, and the stack trace (return addresses) leading to this error is:

411A1C [System.pas][System][@DynArrayClear$qqrrpvpv][35224]

4105E6 [System.pas][System][@FinalizeArray$qqrpvt1ui][32099]

4104A1 [System.pas][System][@FinalizeRecord$qqrpvt1][31777]

4105C5 [System.pas][System][@FinalizeArray$qqrpvt1ui][32058]

E33D58 [STCommon.PankkiUtils.pas][STCommon.PankkiUtils][Pankkiutils.BigIntegerStrMod$qqrx20System.UnicodeStringxi][1083]

E33F01 [STCommon.PankkiUtils.pas][STCommon.PankkiUtils][Pankkiutils.TarkistaRFViite$qqrx20System.UnicodeString][1127]

E3BD9D [STCommon.PankkiUtils.DUnitX.pas][STCommon.PankkiUtils.DUnitX][Pankkiutils.Dunitx.TSTCommonPankkiUtilsTests.TarkistaRFViiteTest$qqrx20System.UnicodeStringxo][47]

4B6810 [System.Rtti.pas][System.Rtti][Rtti.RawInvoke$qqrpvp23System.Rtti.TParamBlock][7750]

4B6C14 [System.Rtti.pas][System.Rtti][Rtti.Invoke$qqrpvx42System.%DynamicArray$18System.Rtti.TValue%24System.Typinfo.TCallConvp24System.Typinfo.TTypeInfooo][7972]

4ADAA5 [System.Rtti.pas][System.Rtti][Rtti.TRttiInstanceMethodEx.DispatchInvoke$qqrrx18System.Rtti.TValuepx18System.Rtti.TValuexi][5883]

4B6FBF [System.Rtti.pas][System.Rtti][Rtti.TRttiMethod.Invoke$qqrp14System.TObjectpx18System.Rtti.TValuexi][9212]

Current memory dump of 256 bytes starting at pointer address 7EF84C30:

00 00 00 00 08 00 00 00 D1 D6 39 E5 57 28 D4 21 5A 80 29 00 00 00 00 00 00 00 00 00 00 00 00 00

00 00 00 00 00 00 00 00 00 00 00 00 80 80 80 80 80 80 80 80 80 80 80 80 00 00 00 00 10 11 F8 7E

00 00 00 00 00 00 00 00 14 BA 43 00 00 00 00 00 66 FD 08 00 BA 9C 40 00 6B C7 40 00 8A CF 40 00

DF 40 5E 00 74 C7 40 00 69 45 41 00 26 3F 5E 00 88 9E 5E 00 B2 9F 5E 00 28 33 41 00 D1 91 5F 00

E8 54 00 00 E8 54 00 00 89 C7 40 00 D5 CF 40 00 94 01 53 00 03 C9 40 00 04 4E 52 00 9C 65 52 00

F1 50 52 00 8B F6 75 00 A0 52 76 00 F6 4E 52 00 22 4A 52 00 34 00 00 00 AC 9A 4F 00 27 55 91 86

34 12 5E 00 00 00 00 00 01 00 00 00 34 0F 5E 00 90 40 5E 00 F0 4C F8 7E 9C 40 5E 00 F0 4C F8 7E

70 E4 FB 7E DC 94 E5 00 00 00 00 00 00 00 00 00 00 00 00 00 D8 AA 6E 79 00 00 00 00 10 11 F8 7E

. . . . . . . . Ñ Ö 9 å W ( Ô ! Z € ) . . . . . . . . . . . . .

. . . . . . . . . . . . € € € € € € € € € € € € . . . . . . ø ~

. . . . . . . . . º C . . . . . f ý . . º œ @ . k Ç @ . Š Ï @ .

ß @ ^ . t Ç @ . i E A . & ? ^ . ˆ ž ^ . ² Ÿ ^ . ( 3 A . Ñ ‘ _ .

è T . . è T . . ‰ Ç @ . Õ Ï @ . ” . S . . É @ . . N R . œ e R .

ñ P R . ‹ ö u .   R v . ö N R . " J R . 4 . . . ¬ š O . ' U ‘ †

4 . ^ . . . . . . . . . 4 . ^ . � @ ^ . ð L ø ~ œ @ ^ . ð L ø ~

p ä û ~ Ü ” å . . . . . . . . . . . . . Ø ª n y . . . . . . ø ~

OK

And then exception

Debugger Exception Notification

Project STCommon.UnitTests.exe raised exception class EInvalidPointer with message 'Invalid pointer operation'.

Break Continue Help

Is My Code broken, is there something I need to do etc...

Velthuis.BigDecimals.pas

LongDateFormat: 'dddd, dd MMMMM yyyy HH:mm:ss';

maybe to

LongDateFormat: 'dddd, dd MMMMM yyyy HH:nn:ss';

Months in time part seems wrong. (If more than less standard Delphi format(s))

using in multithread programs failes due to usage of global unsafe variables

Hello, i test BigInteger and saw a mismatch.
Code

C := GetTickCount;
_BigInteger.RoundingMode := rmRound;
for I := 0 to Iteration do
begin
_Double := I / 100;
_BigInteger := BigInteger.Create(_Double);
_Int64 := Round(_Double);
if _Int64 <> _BigInteger then
begin
S := 'Extended=' + _Double.ToString + ' ,Int64= ' + _Int64.ToString +
' BigInteger=' + _BigInteger.ToString + ' Error';
Memo1.Lines.Add(S);
Break;
end;

end;
C := GetTickCount - C;
Memo1.Lines.Add(C.ToString);

result:

Extended=1,5 ,Int64= 2 BigInteger=1 Error

In bigintegers.pdf

rmRound

Rounds the same way as theSystem.Round function. Any (absolute) fraction > 0.5 (this excludes 0.5 itself) is roundedaway from 0. So 1.5 rounds to 1, but 1.50001 rounds to 2.

But its not equally System.Round (bank rounding).
How Round similar System.Round (bank rounding)?

I'm getting an error when attempting to compile velthuis.bitintegers.primes on the function NextProbablePrime. Errors on BigInteger.Two.

You have a global variable "Two" declared as a BigInteger, and initialize it to 2, and you also have a local variable Two also declared as a BigInteger. But BigInteger class does not have a method Two.

First of all, I would like to commend the brilliant work you have done on this library.
I would also like to ask if there are any future plans to support FreePascal compiler??

Your BigInteger Package has a problem.

On the Velthuis.BigIntegers.Primes unit

// Next probable prime> = N.
Function NextProbablePrime (const N: BigInteger; Precision: Integer): BigInteger;
Var
   Two: BigInteger;
Begin
   Result: = N;
   If Result = BigInteger.Two then
     Exit;
   If Result.IsEven then
     Result: = Result.FlipBit (0);
   While not IsProbablePrime (Result, Precision)
   Begin
     Result: = Result + BigInteger.Two;
   End;
   Result: = N;
End;

[Dcc32 Error] Velthuis.BigIntegers.Primes.pas (191): E2003 Undeclared identifier: 'Two'

I love Bigintegers and BigDecimals!

I can't get BigDecimal.Sqrt to work: always get divide by zero error. Here is my humble attempt:
A := '123.456';
writeln('>> 1 ',A.ToPlainString);
B := BigDecimal.Sqrt(A);
writeln('>> 2');
--- Of course A and B are type BigDecimal. I must be doing something wrong. The same code for BigInteger works fine.

After a little digging:
I changed Veltius.BigDecimals.pas, function BigDecimal.Sqrt(Precision: Integer): BigDecimal;
from:
while (Result * Result - Self).Abs >= Epsilon do
to:
while ((Result * Result - Self).Abs >= Epsilon) and (Result < 0.0)

This seems to work for me.

Hi,
I am writing a faster parser function than the original one and while testing it I raised up an error that may be reproduce with this code:

procedure AParsingError;
var
  N,M : BigInteger;
  R : IRandom;
begin
  R := TDelphiRandom.Create(-332888001);
  N := BigInteger.Create(Random(100)*1000, R);
  BigInteger.Base := 12;
  M := BigInteger.Parse(N.ToString);
  if (N<>M) then raise Exception.Create('Parsed value differs from the original!');
end;

The N.ToString and M.ToString differs per the digits from positions 1983 to 1992 (probabely a Limb).
In attached file uBadValue.zip
there is a unit file embeding the value produced by the random generator (18250 hex-digits long), just in case.

Best regards,
F. MARQUES

SNegativeRadicand contains "%s", so replace
raise EInvalidArgument.Create(SNegativeRadicand);
with
raise EInvalidArgument.CreateFmt(SNegativeRadicand, ErrorInfo);
in
class procedure BigInteger.Error(ErrorCode: TErrorCode; const ErrorInfo: array of const);
begin
case ErrorCode of
ecParse:
raise EConvertError.CreateFmt(SErrorParsingFmt, ErrorInfo);
ecDivbyZero:
raise EZeroDivide.Create(SDivisionByZero);
ecConversion:
raise EConvertError.CreateFmt(SConversionFailedFmt, ErrorInfo);
ecOverflow:
raise EOverflow.Create(SOverflow);
ecInvalidArgFloat:
raise EInvalidArgument.CreateFmt(SInvalidArgumentFloatFmt, ErrorInfo);
ecInvalidBase:
raise EInvalidArgument.Create(SInvalidArgumentBase);
ecInvalidArg:
raise EInvalidArgument.CreateFmt(SInvalidArgumentFmt, ErrorInfo);
ecNoInverse:
raise EInvalidArgument.Create(SNoInverse);
ecNegativeExponent:
raise EInvalidArgument.CreateFmt(SNegativeExponent, ErrorInfo);
ecNegativeRadicand:
raise EInvalidArgument.Create(SNegativeRadicand);
else
raise EInvalidOp.Create(SInvalidOperation);
end;
end;

The function "IsProbablePrime" runs into an endless loop when passing the value "3".
I think the repeat until loop is not completely correct.

Happens if BigDecimal.DefaultPrecision is less then number precision.
If default precision is 64 then BigDecimal.Sqrt of number with 65 or more digits return always 0. "Sqrt(Pi) = 0"

Workaround:
call Sqrt with parameter Max(DefaultPrecision and Number.Precision)

The same problem is for division.
Happens for 10000 digit precison constants i.e. "Pi / 2 = 0"
if Pi decimal places >DefaultPrecision

Maybe it is a feature? :-)

Happens in GetExponent functions only if range checking is on.

Solution:
replace
Result := E - CDoubleBias
with
Result := Integer(E) - Integer(CDoubleBias)
in FloatUtils.pas

function GetExponent(const AValue: Single): Integer; overload;
var
M, E: UInt32;
begin
M := GetRawSignificand(AValue);
E := GetRawExponent(AValue);
if (0 < E) and (E < CSingleExponentMask) then
Result := Integer(E) - Integer(CSingleBias)
else if E = 0 then
if M = 0 then
// +/- Zero
Result := 0
else
// Denormal
Result := 1 - CSingleBias
else
// NaN or +/-Infinity
Result := 0;
end;

function GetExponent(const AValue: Double): Integer; overload;
var
M: UInt64;
E: UInt32;
begin
M := GetRawSignificand(AValue);
E := GetRawExponent(AValue);
if (0 < E) and (E < CDoubleExponentMask) then
Result := Integer(E) - Integer(CDoubleBias)
else if E = 0 then
if M = 0 then
// +/-Zero
Result := 0
else
// Denormal
Result := 1 - CDoubleBias
else
// NaN or +/-Infinity
Result := 0;
end;

function GetExponent(const AValue: Extended): Integer; overload;
var
M: UInt64;
E: UInt32;
begin
M := PUInt64(@avalue)^;
E := GetRawExponent(AValue);
if (0 < E) and (E < CExtendedExponentMask) then
Result := Integer(E) - Integer(CExtendedBias)
else if E = 0 then
if M = 0 then
// +/- Zero
Result := 0
else
// Denormal
Result := 1 - CExtendedBias
else
// NaN or +/-Infinity
Result := 0;
end;

Solution: Please add {$TYPEDADDRESS OFF} to BigIntegers or cast pointers.

How can I use this with C++ Builder 10? The "Velthuis.BigIntegers.hpp" doesn't exist and I can't generate it, since I only have C++ Builder, but not Delphi installed.