Re: AWL Database Cleanup

[Justin Mason <jm...@jmason.org>]

Theo Van Dinter writes:
> On Mon, Apr 28, 2008 at 03:52:02PM -0400, Kris Deugau wrote:
> > There is no way I know of to shrink BerkelyDB files in-place.  >:(
> 
> In case anyone's wondering, this is why the Bayes expire system creates a new
> DB file and copies over the entries that should be kept, then deletes/renames
> the files.
> 
> The AWL DB lacks timestamps and an expiry system.  There were some
> thoughts about making Bayes, AWL, etc, all use a generic system which
> included timestamps for entries and a plugable expiry system, but no
> one ever worked on it. :(

To be honest, at this stage I'd be happy to see just a simple AWL expiry
mechanism -- the over-arching solution sounds like it just got bogged down
in too much generality.  KISS.

--j.

Re: AWL Database Cleanup

[Theo Van Dinter <fe...@apache.org>]

On Tue, Apr 29, 2008 at 09:27:49AM +0100, Justin Mason wrote:
> To be honest, at this stage I'd be happy to see just a simple AWL expiry
> mechanism -- the over-arching solution sounds like it just got bogged down
> in too much generality.  KISS.

Perhaps.  The issue is that it's basically going to be the same thing as the
Bayes system entirely -- versioning, upgrades, magic tokens, new key/value
formats, and expiry system.

Instead of reinventing the wheel, again, and putting in copied/new code
instead of reusing what we already have, I'd think the generic solution is the
better way.

That said, if instead of time-based expiry we wanted to do something else like
count-based expiries, that's doable without having to change everything
around.  The problem starts when we want to change the format of the AWL DB to
do time expiry.

-- 
Randomly Selected Tagline:
"I don't think Microsoft is evil in itself; I just think they make really
 crappy operating systems."            - Linus Torvalds