Hi, [sorry for the delay... darn meetings] On Wed, 26 Jul 2006, Martin Langhoff (CatalystIT) wrote: > Johannes Schindelin wrote: > > > Turns out that DBD::SQLite does not favour preparing statements which > > are never executed. So, turn all 4 statements, which were prepared > > _always_, into methods, like the other 12 prepared statements. > > Can you give me a reference for that "does not favour preparing > statements"? The relevant bug entry is http://rt.cpan.org/Public/Bug/Display.html?id=17603 > I have some vague recollection of timing prepared/unprepared inserts and > getting a huge difference in performance. Yes, preparing before executing is a wonderful thing. But you need to execute the prepared statements; otherwise it is a waste of resources (plus, it does not finalize with DBD::SQLite). > Looking at the source of SQLite.pm doesn't clarify much -- > prepare_cached is actually implemented by DBI, the driver doesn't need > to implement it. Yes. The culprit is in prepare(). > > Now, the only warning left is the gzip one... > > That's harder. I wonder whether using libgit's XS module we can now get > libgit to give us a gzipped file directly. I guess it only has > performance savings for unpacked repos. I still have the problem on at least two of my boxes that Git.xs does not load. The next thing after looking into the gzip thing is to rewrite git-mv in C. Ciao, Dscho - : send the line "unsubscribe git" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html