Re: [Monotone-devel] Re: SQLite3 almagamation

[Matthew Gregan]

At 2007-04-28T11:20:17+0200, Lapo Luchini wrote:
> Anyway I'm interested in performances (both compile time and execution
> speed) so I'll probably try and integrate it ;-) (probably in
> nvm.lapo.sql-amalgamation or something)

It's fairly trivial to integrate, see
org.flim.monotone.experiment.sqlite-amalgamation on venge.net.  The two
immediate concerns I have are:

 * ease of importing future amalgamated SQLite releases (i.e. reviewing
   diffs, especially if we are carrying any local changes)

 * new bugs revealed by the compiler's optimiser, especially since I have
   some vague memory that SQLite needs to be built with
   -fno-strict-aliasing.

I created that branch to see what sort of effect the amalgamation has on the
first of those items.  The second, well, we'll just have to wait and see, I
guess.  I haven't run any benchmarks, so I'd be interested to see if it
shows much difference with monotone.

Cheers,
-mjg
-- 
Matthew Gregan                     |/
                                  /|                    address@hidden