Re: [Monotone-devel] Hash collisions resiliency

[tekHedd]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Nathaniel Smith wrote:
[a comprehensive answer to my question which is exactly what I need to
make me feel comfortable]

Thanks for answering this. I feel much better about the worst case
scenario(*).

It seems like you could run for a while and get some very strange
results, but it's unlikely to result in major data loss. This is not
verified, and definitely should be explored by brave pioneers. However,
the risk of massive data loss feels low.

It's about comfort. Heck, we all know what happens when CVS fails. It's
simple and comfortable. It also really sucks for many distributed
development models.

For those who _yet_ _again_ reminded me how improbable it is when
compared to hardware failure: sure, I'll bite. Perhaps I should be
asking what will happen to my monolithic SQLite database when I have
corruption due to hardware failure, malicious generation of duplicate
keys using as yet undiscovered weaknesses in SHA1(*), or software
bugs(*). ;)

I will have to try this "deliberately crippled hash" trick, although I
was hoping to avoid, you know, work. After I get those taxes done.
Should be a fun weekend.

Thanks much!
tom
(*) - Yes, I know it's nearly impossible.
- --
Tom Surace                )                 ,~~v~~,
address@hidden    )                ,`.   .`,
http://www.byteheaven.com ) ------------- ===  +  === ---
                                       Hamster was here
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFCXwFXDoEO2NB2KMURAiLiAJ9wgSD+hd/J5X4gu1riIfAHCVIeggCgxYu9
9ib38WdG0/iNANuy2poBIWQ=
=bHl7
-----END PGP SIGNATURE-----