> On Mar 27, 2018, at 2:37 PM, Olivier Mascia <o...@integral.be> wrote: > > Been there, changed plans. In our case we profiled no net benefit from such > a pooling compared to have each thread which need a connection to actually > open it, use it and then close it. It looks like you really need a text > heavy / complex schema (taking some noticeable time to parse upon each new > attachment) and a very good connection pool mechanism (not adding too much > contention when obtaining a connection from the pool or returning it) to > really see benefit from this more complicated scheme.
(I think by "attachment" you mean a sqlite3* connection?) Does your program run in any low-memory environments? In my use case (mobile devices) I'm more worried about the RAM consumed by each SQLite connection's cache. (No, we can't use shared-cache mode.) —Jens _______________________________________________ sqlite-users mailing list sqlite-users@mailinglists.sqlite.org http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
