On Thu, 19 Aug 2010 17:15:40 -0400, Tim Romano <tim.romano...@gmail.com> wrote:
>Ah, an opportunity for another purist tirade presents itself. > >I don't have a hack for SQLite but something I consider to be a much better >practice that accomplishes the same goal. If your business rules would >declare that rows with value X in column Y no longer belong to the set, the >most straightforward way to implement such a rule is to move those rows to >another table where they do belong. Use an after update/insert trigger to do >this. [...] >The partial index is one very messy thing, fraught with ambiguities, >something to avoid. I can imagine other business rules being really >bollixed up by the sudden reappearance of zombie rows. +1 My 2 cents: Pure SQL doesn't mind about indexes, in RDBMS implementations they are just an optimization feature and a way to implement unique constraints. Anything more is a can of worms indeed. Optional indexes are a codasyl hierarchical or network database feature, where indexes are exposed to the DML. -- ( Kees Nuyt ) c[_] _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users