Ugochukwu Mmaduekwe

@Geoffrey Smith, thanks a lot for this. I really appreciate. Will test it once I get to my workstation. Thanks once again.

Here is an attempt by someone to implement FPC awesome TProcess in Delphi. May be someone else can work on extending it with Linux support.

Any help please? @Remy Lebeau

Only if you use the web installer. The ISO installer has always contained the Delphi and C++ Builder Installers in one.

Note the key word free in the original poster question.

Have you tried CryptoLib4Pascal ?

Correct me if I am wrong but this will require him to port the headers first of all because the ones I see online are not crossplatform?

Hi Ivana and welcome to the forum. To answer your question, CryptoLib4Pascal is the complete package and more. The hashes you requested and more can be found here sample for generating Key Pairs with Elliptic curves, signing and verifying can be found here and here AES and Chacha20 sample here Perks of CryptoLib4Pascal Constantly developed and improved Cross Platform and Cross Compiler support used in the real world on various products uses the friendly MIT License Best of all, it's free.

NumCPULib4Pascal is a Cross Platform Pascal library to query the number of CPUs (Logical (logical processors) and Physical (cores)) on a machine. Usage Add NumCPULib to uses clause: uses NumCPULib; var lcc, pcc: Int32; begin // count logical cpus (aka logical processors) lcc := TNumCPULib.GetLogicalCPUCount(); // count physical cpus (aka cores) pcc := TNumCPULib.GetPhysicalCPUCount(); end; What is the difference between the existing System.CPUCount and NumCPULib4Pascal? 1. System.CPUCount only reports the LogicalCPU Count (aka logical processors), it has no option to report the PhysicalCPU Count (cores). 2. System.CPUCount will not report the correct value on windows systems with more than 64 logical processors. NumCPULib4Pascal fixes this by querying GetLogicalProcessorInformationEx on these OSes. GitHub Repository NumCPULib4Pascal

if you use Berlin upwards, you can use the inbuilt version found here else you can try the code below. you will need this library though program SimpleHMAC; uses SysUtils, HlpHashFactory; var param_str, secret, sign: string; begin try secret := 't7T0YlFnYXk0Fx3JswQsDrViLg1Gh3DUU5Mr'; param_str := 'api_key=B2Rou0PLPpGqcU0Vu2&leverage=100&symbol=BTCUSD&timestamp=1542434791000'; sign := THashFactory.THMAC.CreateHMAC(THashFactory.TCrypto.CreateSHA2_256(), TEncoding.UTF8.GetBytes(secret)).ComputeString(param_str, TEncoding.UTF8).ToString(); WriteLn(LowerCase(sign)); ReadLn; except raise; end; end.

Well, have you tried Mormot's implementation? https://github.com/synopse/SynPDF/blob/master/SynCrypto.pas You can also try HashLib4Pascal if you want. https://github.com/Xor-el/HashLib4Pascal/ Also you can try out Wolfgang Ehrhardt aka Gammatester (RIP) excellent Hash libraries which contains HMAC implementations too. http://www.wolfgang-ehrhardt.de/crchash_en.html These are guaranteed to work on Delphi 2010 and above and don't require any external DLLs for as long as I remember.

@Clément, To be sincere, I tend to keep pointer constructs to a minimum (to be on the safe side) since the size of integers varies on a various platforms supported by Delphi. Also any reason why you reimplemented HMAC when it is supported in modern Delphi versions?

Well I use Lazarus on Windows and Linux and it is quite stable (especially on Linux), I think the issue you are experiencing may be related to just Mac OSX. Can you try to install the trunk version using fpcdeluxe? Your issue might have been fixed there.

32 or 64bit Lazarus? Trunk or stable release?

Inline is a way to hint the compiler to avoid calling a method and instead take the contents of that method and place it at the call site.

This might have something to do with newer OpenSSL Dlls switching from using the Legacy CryptoAPI to using BCryptGenRandom for generating OS Random numbers.

Lol, at least we now agree on a common fact that for now Delphi doesn't have a PurePascal EdDSA implementation.

Seems you have made up your mind that nothing good and optimal can come out of non native code. In this case, there is very little I can say to convince you otherwise. By the way, the current media player I use is written in JavaScript and it runs beautifully for me on both my Windows and Linux boxes.

Really? Do we have to go down this road again? ECDSA (Elliptical Curve DSA) <> EdDSA (Edwards Curve DSA). It has a lot of use cases as ECDSA and it is been used in Monero and Nano to mention but a few.

Really? I have seen highly performant programs including media players written in .NET and even JavaScript.

This implements just ECDSA using only secp256r1 curve.

@Joseph MItzen you just analysed everything perfectly from my point of view. 👍

Yes It is my project and no it doesn't implement Edwards curves.

Asymmetric Cryptography especially stuffs involving elliptical curves like Edwards curves.

Unfortunately, the CodeTyphoon team seems more focused on the IDE than the compiler.