On Thu, Sep 01, 2011 at 07:41:26PM +0200, Florian Weimer wrote:
> Please try again with the -s option:
> 
> | If the database contains multiple databases and the -s flag is not
> | specified, the statistics are for the internal database that
> | describes the other databases the file contains, and not for the
> | file as a whole.

Okay, that works.  Somehow I missed that.  db5.1_stat should probably be
a little clearer that the information is for the database of databases
and not for the single named database inside it.  Consider this a
minor bug for that purpose, then.

> Is there any particular reasony why you want to use hash databases?
> They are rarely better than B-tree databases.

The documentation specifies that hash databases perform better on large
datasets where there is poor locality of reference.  I have a large
dataset and the first 32 bytes of the 64-byte keys are an SHA-256 hash
of the individual objects which are then composed into larger objects.
This does not have good locality of reference, so a hash DB is a better
choice.

-- 
brian m. carlson / brian with sandals: Houston, Texas, US
+1 832 623 2791 | http://www.crustytoothpaste.net/~bmc | My opinion only
OpenPGP: RSA v4 4096b: 88AC E9B2 9196 305B A994 7552 F1BA 225C 0223 B187

Attachment: signature.asc
Description: Digital signature

Reply via email to