Re: [PLUG] free courses, systemd Win, hashquines

[brent timothy saner via plug <plug@lists.phillylinux.org>]

On 9/24/22 16:30, Walt Mankowski via plug wrote:
> On Sat, Sep 24, 2022 at 02:30:46PM -0400, brent saner via plug wrote:
> > MD5 is well and truly broken. It shouldn't be trusted for anything except
> > transmission integrity checking, and even then there are significantly
> > better alternatives.
>
> I don't trust MD5 for external files, but I still run md5sum on
> directories to check for duplicate files, mainly because it's a lot
> faster than sha1sum. Are there alternatives you recommend?
> ___________________________________________________________________________
> Philadelphia Linux Users Group         --        http://www.phillylinux.org
> Announcements - http://lists.phillylinux.org/mailman/listinfo/plug-announce
> General Discussion  --   http://lists.phillylinux.org/mailman/listinfo/plug

If you have *total control* over the directory/directories, that's still 
a valid use case for MD5, IMHO.

But! If you have the *option*, I'd highly recommend BLAKE2[0]. (There is 
a BLAKE3, which is even faster, but support/implementation/availability 
is far from prevalent.)

Like MD5 (and a handful of others), it has its own RFC.[1]

There are two basic variants, BLAKE2s and BLAKE2b. You likely want 
BLAKE2b, where given the choice; BLAKE2s is moreso for 32-bit machines.

It's faster* than MD5 for checksumming (what you'd usually use it for 
anyways these days) with the benefit of being "as strong" as the SHA-3 
suite for integrity. It uses ChaCha under the hood, which - to my 
knowledge - has no vulns (but CAN be extremely tricky to *use* without 
any vulns if the spec isn't followed exactly).

Mini-Rosetta for you:

(SH) md5sum => b2sum (part of coreutils; it's likely already installed)

(Perl) Digest::MD5 => https://metacpan.org/pod/Digest::BLAKE2 (and 
likely others; I'd imagine there's a pure Perl implementation)

(Python) hashlib.md5() (unless FIPS) => hashlib.blake2b()/hashlib.blake2s()

(Golang) crypto/md5 => golang.org/x/crypto/blake2b, 
golang.org/x/crypto/blake2s

(Rust) md5 => blake2

HTH



* Basic benchmarks show it roughly 2.5% faster than MD5 for BLAKE2s, and 
a quite impressive ~33% faster than MD5 for BLAKE2b.


[0] https://www.blake2.net/
[1] https://www.rfc-editor.org/rfc/rfc7693
___________________________________________________________________________
Philadelphia Linux Users Group         --        http://www.phillylinux.org
Announcements - http://lists.phillylinux.org/mailman/listinfo/plug-announce
General Discussion  --   http://lists.phillylinux.org/mailman/listinfo/plug