Re: [Monotone-devel] Re: summit flag day?

[Richard Levitte]

In message <address@hidden> on Wed, 23 Apr 2008 20:15:58 +0200, Lapo Luchini 
<address@hidden> said:

lapo> Koen Kooi wrote:
lapo> > Please ensure that new mtn releases can still talk to 'old' releases 
over netsync
lapo> 
lapo> Some of the changes I mentioned can allow this, some other simply do 
not...
lapo> 
lapo> As far as I know netsync (i.e. very little) changing the hash to
lapo> something that's not 20 byte long will necessarily break it. The
lapo> same can probably be said about changing certificate format. And
lapo> changing the way keys are referred to probably also.

- we can have variable length hashes, where 20 bytes with no algorithm
  specifier is assumed to be SHA-1.  This has been done before...
- I suspect that we can add an optional capability specification in
  netsync, it would probably be an extension of some command (that's a
  pure guess, I haven't looked yet, but I've seen this sort of thing
  done before)

The only thing that I see might need a flag day of all those things is
this mega-cert change, which I'm not sure is needed, relatively
speaking.

lapo> The new certificate format should IMHO contain at least one
lapo> 'version' byte, which also defines the hash algorithm used, so
lapo> that he can be more future-proof (at the price of a single
lapo> byte).

I'd rather go with a more flexible aproach for algorithm specifiers,
prepending the signature with something like '{SHA1}'.  Yes, it's more
expensive on disk, but previous experience has shown that it's usually
safer.

Cheers,
Richard

-----
Please consider sponsoring my work on free software.
See http://www.free.lp.se/sponsoring.html for details.

-- 
Richard Levitte                         address@hidden
                                        http://richard.levitte.org/

"When I became a man I put away childish things, including
 the fear of childishness and the desire to be very grown up."
                                                -- C.S. Lewis