> On Nov 21, 2017, at 1:56 AM, R Smith <rsm...@rsweb.co.za> wrote: > > That assumes you are not starting from an integer part (like 4000) and > hitting the exact same relative insert spot every time, which /can/ happen, > but is hugely unlikely.
Not to beat this into the ground, but: it’s not that unlikely. Let’s say you sort rows by date. You’ve already got some entries from 2015 in your database, and some from 2017. Someone now inserts 60 entries from 2016, and to be ‘helpful’, they insert them in chronological order. Wham, this immediately hits that case. (This is similar to the problem that some tree data structures have, where adding entries in sorted order results in the must unbalanced possible tree.) It’s a lot better to use strings, and just increase the length of the string as necessary. So to insert in between “A” and “C” you add “B”, then to insert between “A” and “B” you add “AM”, etc. —Jens _______________________________________________ sqlite-users mailing list sqlite-users@mailinglists.sqlite.org http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
