On Sat, Sep 11, 2010 at 8:54 PM, Simon Slavin <slav...@bigfraud.org> wrote:
> > On 11 Sep 2010, at 5:38pm, Stephen Oberholtzer wrote: > > > It would be an interesting attempt to try efficiently patching SQLite > > to recognize this situation and read data directly out of the index. > > But I've been told many times that SQLite already does this. > Simon, I think this is because the index is forced when you use ORDER BY so if the index matches it, this is the only way to do the retrieval effectively. But when you tell you want some field(s) without order by and they're also contains in the index, sqlite optimizer would recognize there would be fewer reads in case of the index, but I suppose it currently doesn't do this. I think there might be some indirect logic behind this or maybe this thing with size difference not so obvious. For example if the field mentioned is TEXT and it's much more larger than all other fields together,the benefits of using the index for retrieval are not so great, although index at least should not be worse in any situation Max _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
