[darcs-users] Why Bitkeeper still wins

[Jamie Webb]

On Tue, Mar 22, 2005 at 08:20:48AM +0100, Peter Busser wrote:
> And it uses 
> SHA-1 hashes for everything. SHA-1 is starting to reach the end its useful 
> life.

That depends on your definition of useful. The break in SHA1 is
largely theoretical at this point. No-one has successfully been able
to manufacture a hash-preserving change to an existing file, and it's
likely to be some time before they do. It's a vastly harder problem,
and monotone will almost certainly have moved to a different hash
function before it's an issue.

Regardless, SHA1's cryptographic weaknesses are /completely/
irrelevant to it's performance as a hash function for benign inputs,
so Monotone is certainly no worse off than Darcs in this respect. It
concerns me very very slightly that Monotone doesn't use any sort of
global namespace to identify patches (as Darcs uses email addresses),
but Monotone does allow one to make the collision probability
arbitrarily small using some sort of grouping scheme, so again this is
strictly a theoretical problem.

I'd say the binary formats and relatively poor merging ability are
by far the stronger arguments against monotone.

-- Jamie Webb