Dominique Devienne wrote: > On Mon, Jun 11, 2018 at 4:27 PM Clemens Ladisch <clem...@ladisch.de> wrote: >> It does write to the same pages, but those pages must be copied to the >> rollback journal so that they can be restored if the transaction is >> rolled back. (Or are the two passes inside the same transaction?) > > They are part of the same transactions. So the no-rewrite pages should > still apply, no?
Indeed. I'd guess that just writing all those zero-filled pages is too much overhead. And the incremental blob I/O function are intended to access partial blobs; when you write the entire value at once, just use SQL. >> Don't insert the zero blobs in the first pass; this still rewrites all >> rows, but there is less data that might need to be rolled back. >> >> Consider moving the blobs to a separate table. > > Is that really/still relevant, considering the above? Well, there's only one way to find out ... ;-) Regards, Clemens _______________________________________________ sqlite-users mailing list sqlite-users@mailinglists.sqlite.org http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users