Zooko:

Yes, the new version will be open sourced.  I am retired and living my
dream of doing research on high speed data structures that support the
things that Judy does.  I want my work to be public and free forever.
My goal is to make it so fast and efficient that even hash methods
cannot measure up to the speed of Judy.  I do not think it will be very
long before I will have new prototype versions of Judy available for
inspection and testing.  My current design is somewhat different then
the published Judy.  It takes far more advantage of the CPU Cache.  I am
also trying to write it so that it is easier to read and understand than
the current Judy.  It will be interesting to see if people measure it as
faster than hashing methods.  My current measurements are very
encouraging.  I hope I get a lot of feedback and help with the API.
Some things, such as Next/Prev will be many times faster than the
current Judy.

A comment on your project.  One thing I have learned over and over again
is:  there are always better ways to do things even when you think you
got the best way possible designed.  I remember thinking about how to
take better advantage of the cache I was not using on the current Judy
back in 2001.  I now feel very naive, because it was a fairly
simple/obvious thing to do.  Most of the things in the new Judy design
will appear as "that's a fairly obvious way of doing it".  But it took
me years to think of it -- stupid me, I was asking the wrong questions.

Thanks for you interest,

Doug


Doug Baskins <[email protected]>



----- Original Message ----
From: zooko <[email protected]>
To: Doug Baskins <[email protected]>
Cc: [email protected]
Sent: Sunday, February 15, 2009 10:12:03 PM
Subject: Re: AW: AW: New Judy version

Dear Doug Baskins:

Are you planning to open-source your new version of Judy?

If so, a good practice is to go ahead and publish what you've got,  
even if it doesn't compile, is ugly, doesn't really show what you  
mean, and so on.

This practice of "release early, release often" has a lot of good  
effects, one of which is to assure people that the ultimate result,  
when it arrives, will be open-source.

This sort of thing could be important for some long-term plans.  For  
example, I have a scheme in mind for a new data structure, inspired  
by the success of JudyTrees but probably quite different in its  
design.  If I'm sure that your next project will be open source, then  
I think of you as a partner -- two different teams exploring adjacent  
areas of the design space.

If I think that you (or someone) might assert exclusive rights over  
your next project, then I think of that project as a potentially  
dangerous competitor to mine, and I want to withhold information from  
you.

My own project is in such a nascent stage that it is hardly worth  
mentioning and I wouldn't want to waste your time with it until I've  
done some actual experiments so I can report real numbers.  But I  
thought I would mention it so as to encourage you to "work in public"  
by publishing your incomplete work, in order to elicit such  
cooperation from people other than me who may already have more to  
offer.

Regards,

Zooko Wilcox-O'Hearn

------------------------------------------------------------------------------
Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco, CA
-OSBC tackles the biggest issue in open source: Open Sourcing the Enterprise
-Strategies to boost innovation and cut costs with open source participation
-Receive a $600 discount off the registration fee with the source code: SFAD
http://p.sf.net/sfu/XcvMzF8H
_______________________________________________
Judy-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/judy-devel


------------------------------------------------------------------------------
Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco, CA
-OSBC tackles the biggest issue in open source: Open Sourcing the Enterprise
-Strategies to boost innovation and cut costs with open source participation
-Receive a $600 discount off the registration fee with the source code: SFAD
http://p.sf.net/sfu/XcvMzF8H
_______________________________________________
Judy-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/judy-devel

Reply via email to