Information générale
Catégorie:
Base de données, Tables, Vues, Index et syntaxe SQL
David,
I surely agree that the factors you name cause bloat. I even saw this in the limited test I ran.
It's also true that I jumped the gun when I simply ASSuMEd that a "single pass" reindex would be much much quicker than a rebuild. I have no idea what REINDEX now does internally. But I have come to infer that VFPs natural buffering (cacheing) can often give the same benefit as the "single pass" method.
As regards when the bloating caused by REINDEX disappeared, I suppose that it really is immaterial. I do find it disturbing that we seem to have to "stumble upon" discovery of changes to long accepted 'axioms' for VFP coding/feature usage. But that is another issue.
Cheers,
JimN
>Jim,
>
>Most CDX bloat is caused by the nonrecovery of space used by index tags that have been deleted. A lesser bloat is caused by B+ tree fragmentation caused by the normal insertion of records. I don't recall when the REINDEX command started recovering deleted tag space. I don't think it did back in FP2.x. It does in VFP5 and VFP6 which are the only versions I have on this machine. It is also quite possible that REINDEX effectively is doing 1) read the indexes, 2) delete tag all, issuing the individual index commands
>
SNIP
Précédent
Répondre
Voir le fil de ce thread
Voir le fil de ce thread à partir de ce message seulement
Voir tous les messages de ce thread
Voir tous les messages de ce thread à partir de ce message seulement