Paul McCullagh wrote:
On May 9, 2009, at 3:15 AM, Monty Taylor wrote:
Couple of quick followups. (On a phone, so a bit tersly)
Very happy about the interface. Moving forward, I'm actually more
inclined to agree with brian about getting rid of the darned things
and moving them to the field objects,
This will only work if we have field objects that are private to the
handler.
It will also force us to work on the sets on a bit-for-bit basis.
Bit maps, on the other hand, can calculate set operations (union,
intersection, etc), very efficiently.
Not necessarily. The interface can add support for union, subset,
intersection quite easily. The implementation can be efficient
underneath. Mats' bitvector is a good example of that. It just happens
that the std::bitset doesn't have an efficient implementation at this
point for these operations...
Brian's point about encapsulating behind the interfaces means we can
switch out and refactor the implementation without blowing up half the
code base... :)
but I will hack up a vector possibility for testing purposes. I also
want to figure out if we can make use of dynamic_bitset without
getting into a world of build hell.
I've got step one of hudson performance integration done. Jay and I
will finish it this weekend.
Thanks everyone!
Monty
--
Paul McCullagh
PrimeBase Technologies
www.primebase.org
www.blobstreaming.org
pbxt.blogspot.com
_______________________________________________
Mailing list: https://launchpad.net/~drizzle-discuss
Post to : drizzle-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~drizzle-discuss
More help : https://help.launchpad.net/ListHelp
_______________________________________________
Mailing list: https://launchpad.net/~drizzle-discuss
Post to : drizzle-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~drizzle-discuss
More help : https://help.launchpad.net/ListHelp
