[mythtv-users] Database internals

[Paul Gardiner]

On 03/01/2014 00:03, Jay Ashworth wrote:
> ----- Original Message -----
>> From: "Paul Gardiner" <lists at glidos.net>
>
>> Another thing still bugging me, is why mysqldump doesn't create
>> exact copies of the database. I'd have thought that was what it
>> was intended to do.
>
> Having spent hours with mysqldump, I'm going to take a swing at this one:
>
> What the words "exact copy of the database" actually means to you is
> something that actually varies quite a lot.
>
> It can't mean "a binary copy of the internal files" -- aside from endianness,
> you *don't want to know* how much difference there can be amongst different
> DB engines storing the same data.
>
> In point of fact, *the reason SQL exists* is precisely to insulate users and
> programmers from that.

Yeah, good point: "exact copy" clearly wasn't a good phrase. So let's
say a copy that cannot be distinguished from the original by any
sequence of updates and queries. If a copy mechanism existed satisfying
that criteria then it would certainly be safe to transfer or backup a
mythconverg database, because otherwise the queries and updates
performed by mythtv could be used to distinguish the copy from the
original. My guess was that mysqldump with default options created
such a copy. I don't know that, but surely it's possible to do, and
it would be the natural default behaviour.

Cheers,
	Paul.