[Xapian-discuss] Two questions

[Olly Betts]

On Sun, May 15, 2005 at 10:47:53PM +0100, Olly Betts wrote:
> On Fri, May 13, 2005 at 10:17:28AM +0200, roki roki wrote:
> > When I generate first time database with
> > this script and then execute quartzcompact I get the following result:
> > 
> > postlist: Reduced by 49.7817% 912K (1832K -> 920K)
> > termlist: Reduced by 39.4366% 224K (568K -> 344K)

Some progress.  I changed how the keys are built for the termlist table
so appending new documents always appends entries to the end of the
termlist table.  This improved on the above and the termlist is now 424K
(so compaction would give a 19% reduction).  That's much more like it
though we should be able to do better I think.

> > When I execute 5 time this test script and the do quartzcompact I get:
> > 
> > postlist: Reduced by 74.8908% 2744K (3664K -> 920K)
> > termlist: Reduced by 69.7183% 792K (1136K -> 344K)

And this is now 1072K for the termlist (so compaction would give
a 68% reduction).  Not a great improvement, but a move in the right
direction at least.

I've also improved keys for record, value, and position tables in
a similar way.

Unfortunately these changes require a database format change, but I've
put them in flint.

Cheers,
    Olly