[Schevo-devel] Scalability

[Patrick K. O'Brien]

David Binger wrote:
> 
> On Dec 7, 2005, at 1:03 PM, Patrick K. O'Brien wrote:
> 
>> If I had my way I'd try to get one of the Durus folks to create a  custom
>> version of Durus that was optimized for Schevo.  I figure there are
>> opportunities there because the way Schevo stores objects in Durus is
>> unique, regular, and completely different than the typical Durus
>> use-case (just to redundantly emphasize our uniqueness).
> 
> Would you mind describing the schevo-specific need and the potential 
> advantage?

Don't mind at all, except that I haven't really given it a lot of
thought or attention.  It just seems to me that with the various BTree
options that Durus exposes (and that we've yet to experiment with) and
all the indexes we store in Durus that there *must* be some room for
optimization.  Of course, all the gory details are available here:

http://schevo.org/schevo/database.py.html

What I would love is for some Durus expect (hint, hint, cough, cough) to
look at that code to see how we are using (and, perhaps, manhandling)
Durus and tell us "hey, you idiots, don't do it that way" or "hey, you
idiots, here is a way that would be faster and better" or "hey, you
geniuses, you've invented the absolute best way to get the most out of
Durus".  I'd be happy with any of those.  :-)

-- 
Patrick K. O'Brien
Orbtech       http://www.orbtech.com
Schevo        http://www.schevo.org