[merge] 1ms resolution for commit timestamps

John Arbash Meinel john at arbash-meinel.com
Tue Jul 18 20:28:07 BST 2006


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

Aaron Bentley wrote:
> John Arbash Meinel wrote:
>>> Anyway, 1ns is a little excessive, especially because we don't actually
>>> have that much precision, but restricting it to 1s was also too limiting.
> 
> What would be the problem with having multiple commits with the same
> timestamp?
> 
> Aaron

No specific problem with it. Just that we have more information than
that. So it doesn't really need to be thrown away.
1ms seemed like a decent compromise, 1s seemed too restrictive, and not
restricting at all seemed to open us up to floating point precision
bugs. We are lucky that timestamps are >1Billion, if we were closer to
1970, our timestamps would fail to round-trip properly, because the
precision of a double is generally 16 digits.

John
=:->
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2.2 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFEvTZHJdeBCYSNAAMRAiXrAKDA8K+/D1H/Bbuv1MGSa/2yBI5zsgCgnovP
X7HaaB8MzBcZ3X/2nB4KVes=
=dicv
-----END PGP SIGNATURE-----




More information about the bazaar mailing list