On Mon, 18 Dec 2017 07:21:50 -0700 (MST) Dinu <dinumar...@gmail.com> escribió:
> Hick Gunter wrote > > If you are running an SQLite version 3.11.0 or newer, only pages that are > > changed for the first time in the transaction are copied to the WAL file, > > thus extending it. > > > > While progressing the transaction, the rate of "newly changed" to "changed > > again" pages will shift towards re-reading and rewriting the WAL file copy > > of a page; this will tend to slow down the rate at which the WAL file is > > growing, even at a constant delete rate, until every page has been updated > > at least once, and then stop growing until all the remaining deletes have > > been processed. > > Running 3.19; thanks for the explanation, this never occured to me and makes > perfect sense. > Is it compiled with SQLITE_ENABLE_UPDATE_DELETE_LIMIT option? If yes you can delete, let's say 100000 (10E5) rows each round. Index and metadata would fit in sqlite cache and clean wal file, making it faster than a on big delete. -- Eduardo <eduardo.mor...@mobelservices.com> _______________________________________________ sqlite-users mailing list sqlite-users@mailinglists.sqlite.org http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users