Quoting Bennett Todd (bet@newritz.mordor.net): > When used with most normal filesystems, the performance of this database falls > off savagely as the number of records in a table grows past a few thousand, > unless the hashdepth is set appropriately. And most normal filesystems won't > tolerate excessively large keys. I would like to reinforce that; this module will be effectively unusable with a normal filesystem for any significantly large database. The "value" files will be rounded up to the nearest blocksize, wasting absolutely epic amounts of disk space, and a standard UNIX filesystem will run out of inodes pretty quickly. On the other hand, reiserfs (see http://www.idiom.com/~beverly/reiserfs.html to get an idea; find a more recent mirror if you actually intend to use it) is rockingly good for this sort of thing. So good, in fact, that it renders the hashdepth feature redundant. I'm hoping it will go into the standard Linux kernel soon. -- Adam Rice -- wysiwyg@glympton.airtime.co.uk -- Blackburn, Lancashire, England ==== Want to unsubscribe from this list? (Don't you love us anymore?) ==== Well, if you insist... Send mail with body "unsubscribe" to ==== fwp-request@technofile.org