- From: Jonas Sicking <jonas@sicking.cc>
- Date: Mon, 9 Nov 2009 09:00:43 -0800
- To: Anne van Kesteren <annevk@opera.com>
- Cc: Maciej Stachowiak <mjs@apple.com>, Arthur Barstow <art.barstow@nokia.com>, Charles McCathieNevile <chaals@opera.com>, public-webapps <public-webapps@w3.org>
On Mon, Nov 9, 2009 at 3:51 AM, Anne van Kesteren <annevk@opera.com> wrote: > On Mon, 09 Nov 2009 08:12:22 +0100, Jonas Sicking <jonas@sicking.cc> wrote: >> >> * SQL doesn't give any performance guarantees. Many times people tweak >> their SQL in order to get the implementation to use a desired >> evaluation stategy. This won't work in the likely event that different >> implementations use different evaluation strategies for the same >> query. > > From what I understood so far this would also be the case with the Web > B-Tree Database proposal (maybe even more so given that all SQL > implementations currently have the same underlying engine). Am I missing > something? I don't think this is the case with the SimpleDB proposal. I think it's possible to for example guarantee that a lookup based on an index is O(log n) on the number of items in that entity store. Nikunj, is this correct? Anne, are there any specific operations you have in mind where you don't think we can give timing guarantees? / Jonas
Received on Monday, 9 November 2009 17:01:38 UTC