Re: AW: AW: How to check integrity of database?

On Wed, 2005-10-05 at 09:38 -0500, Matt England wrote:
> At 10/5/2005 08:27 AM, Gary Thomas wrote:
> >I [arguably] don't know all the details of how it works, but these sort
> >of checks are definitely a part of what happens. For example, I took
> >one of my FSFS repositories and munged one of the internal files (I
> >changed a single byte, just like your example). When verify got to
> >that revision, I got a nice error:
> > ...
> > * Verified revision 99.
> > svn: Checksum mismatch while reading representation:
> > expected: 4ec62b6b4fd5336a67ce148a9b4fabd7
> > actual: c200993287dccf718c8ba2bdb5075f25
> >which told me that it had failed trying to verify revision 100 (the one
> >I messed with).
>
> Ok, so it looks like there's some hashing/checksuming going on to verify
> file content integrity, judging by the above output at least....unless this
> checksum is only on the meta data, and not the actual file content data?

There's no such distinction (as far as I can tell). What's kept
are representations of the "revision" which contains information
from how to get from one revision to another. The actual file(s)
do not have a separate representation within the native file system.

-- 
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org