yhash not supported in Windows XP about yori HOT 4 CLOSED

At its core, Yori is still just a shell around functionality provided by a platform. Trying to move into the territory of implementing platform functionality would be a substantial extension, and it's not something that one person can really do by themselves.

Frankly, I'd be a terrible person to implement a crypto library. The real task here is taking an existing crypto library, and backporting it to run on an older compiler, operating system, and CPU. After that it still needs to be maintained, since crypto is an eternally evolving space.

Note that the core of the BCrypt.dll engine, which Yori is using, is now open sourced at https://github.com/microsoft/SymCrypt . It unfortunately doesn't directly expose the same interfaces as BCrypt.dll, but from Yori's point of view, the ideal thing would be to take that code, add BCrypt.dll exports, do all of the backporting, and end up with a drop-in BCrypt.dll for older systems. That not only enables Yori to work, but also enables other software that wants newer Windows crypto to work. That also means it's completely decoupled from any of the code here.

Agree that I should update the help text to indicate SHA1 is used by default.

from yori.

why not using Wincrypt? I think XP SP3 with CSP update capable of SHA-512 (not bad)

from yori.

With a couple of syntax changes to accommodate Open Watcom 2.0, I was able to build and run Microsoft's own example code for calculating MD5 checksums:

https://docs.microsoft.com/en-us/windows/win32/seccrypto/example-c-program--creating-an-md-5-hash-from-file-content

The binary seems to run fine under XP SP3 and NT 4.0 SP6.

NT 3.51 is too old and complains that the CryptAcquireContextA procedure is missing from ADVAPI32.DLL.

If you're using MSVC 2.0 then I suspect wincrypt.h is missing from the headers but maybe that's not too difficult to work around and conceivably a lot easier than wrestling with BCrypt.

(Obviously MD5 isn't perfect, but it's still useful.)

from yori.

Thanks, I didn't know these functions existed. Agree that if there's an option that exists on more versions, Yori should use it.

It doesn't matter much about working around missing header defines or exports. That already needed to exist for the Vista+ version anyway, so it's just about changing from one set of functions to another.

I have changes pending for this which I'll push soon. As you've already noted, MD5 and SHA1 exists on NT4+. SHA256/384/512 exists on XP SP3 and 2003 SP2. There's some hacking around requesting the "enhanced" provider which provides those hashes, but it's not too serious.

from yori.