When I modified Sqlite to block type conversions my memory of it is that the changes were quite simple and easy to implement. You could log the change events with a handful of patches to Sqlite. It would be an interesting exercise to analyze the impact of the conversions.
On an aside I discovered a long time ago that analyzing radix changes in commercial type software indicated an enormous overhead which was fairly easily removed by holding numbers in display format whereever possible. Now faster processors make such optimizations less significant. Hardy, Andrew wrote: > Is there any way to log conversions to highlite any issues that would > have been hilighted by failure with strict typing? > > It is my plan to match the column type & data stored type, but clearly > if any conversions are occurring I will be unaware. > > Kind Regards > > > -----Original Message----- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On Behalf Of Dennis Cote > Sent: 27 August 2008 15:45 > To: General Discussion of SQLite Database > Subject: Re: [sqlite] Manifest Typing performance impact? > > Hardy, Andrew wrote: > >> >>Is there a performance hit assosiated with manifest typing? >> > > > I'm sure there is since sqlite must track the type of each data item as > well as its value. But in reality this overhead is quite small. > > >>Is it right that although info on the sqlite site suggests there is an > > >>avilable mode that supports strict typring, this is not infact the > > case? > >> > > > That is correct, the strict affinity mode does not exist. > > >>If there is a performance hit, what are the best ways to minimise > > this? > >>And is there any way to at least log conversions to highlite any >>issues that would have been hilighted by failure with strict typing? >> > > > The best way to minimize the conversion overhead is to store the data in > suitably typed columns (i.e. that match the type of the data stored in > the column). This will avoid any unnecessary conversions when storing, > loading, or comparing the values. > > See http://www.sqlite.org/datatype3.html for the column type affinity > deduction rules. > > HTH > Dennis Cote > _______________________________________________ > sqlite-users mailing list > [email protected] > http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users > _______________________________________________ > sqlite-users mailing list > [email protected] > http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users _______________________________________________ sqlite-users mailing list [email protected] http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
