On Tue, 2007-11-27 at 23:30 +0000, John A Meinel wrote:
> Are we assuming that if there is a name conflict then the packs are
> identical?
No. The existing pack could be corrupt (io error in the past), or
malicious (collision, or deliberately something that doesn't match the
md5sum as its name).
> Or do we also need to grab at least the indexes and make sure that they
> have identical data.
We should check the pack and the indices. The indices may vary because
we haven't solved the 'indices must be regeneratable' bug yet.
On Tue, 2007-11-27 at 23:30 +0000, John A Meinel wrote:
> Are we assuming that if there is a name conflict then the packs are
> identical?
No. The existing pack could be corrupt (io error in the past), or
malicious (collision, or deliberately something that doesn't match the
md5sum as its name).
> Or do we also need to grab at least the indexes and make sure that they
> have identical data.
We should check the pack and the indices. The indices may vary because
we haven't solved the 'indices must be regeneratable' bug yet.
The www.robertcolli ns.net/ keys.txt>.
--
GPG key available at: <http://