Thanks Dominique, Thanks Simon, Do you mean to say SQLite might keep function results across queries? My design would be more complicated, but it is something like this: If my function uses first part of a text column in the row involved and if I make sure all modifications are always appended to the cell, then the function will always return the same value. So it can be deterministic and would work even if SQLite caches function results across queries. Am I correct?
---- On Mon, 18 Feb 2019 19:18:21 +0530 Simon Slavin <[email protected]> wrote ---- > On 18 Feb 2019, at 1:15pm, Arun - Siara Logics (cc) <[email protected]> wrote: > > > Thank you, for the detailed advice, info and the pointer. Is there a > > faster way to query the table using row id, that is, skip the query > > parsing and planner? > > No. For fastest queries, use "WHERE rowid = <whatever>", and list the > columns you're interested in specifically. Do not use "SELECT *". > > > I still need the page cache feature and allow for concurrent modification > > of the row involved, while ensuring determinism by designing so. I guess > > sqlite3_exec() would take care of this, but is there a faster way? > > If you're allowing for the table row you're reading to be modified, then how > will your function be deterministic ? Would changing values in that row not > lead to a change in the value returned by your function ? If not, why are > you looking up the row ? > > Note that if you mark that function as deterministic, you cannot rely on > SQLite calling your function at all. SQLite may reason "I called that > function, with those arguments, a few instructions ago, so I already know > what the result will be.". > > Simon. > _______________________________________________ > sqlite-users mailing list > [email protected] > http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users > _______________________________________________ sqlite-users mailing list [email protected] http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
