Comments on: Smallest Useful Attribute System Part 2 http://meshula.net/wordpress/?p=174 Spaces Between Mon, 30 Aug 2010 08:23:36 -0400 http://wordpress.org/?v=2.9.2 hourly 1 By: admin http://meshula.net/wordpress/?p=174&cpage=1#comment-6454 admin Fri, 27 Jun 2008 06:47:02 +0000 http://meshula.net/wordpress/?p=174#comment-6454 I agree with your point about the maps not changing, although I went down a different path - I have a new version almost ready to post. I ended up using a Ternary Search Tree, as described here: http://www.cs.princeton.edu/~rs/strings/ It gets rid of string compares for lookup, so finding an attribute is really quick; search is on the order of the length of the string you're looking for, instead of log2(n) of the number of strings. Strcmp no longer shows up on my profile, which makes me very happy. I agree with your point about the maps not changing, although I went down a different path – I have a new version almost ready to post. I ended up using a Ternary Search Tree, as described here:

http://www.cs.princeton.edu/~rs/strings/

It gets rid of string compares for lookup, so finding an attribute is really quick; search is on the order of the length of the string you’re looking for, instead of log2(n) of the number of strings. Strcmp no longer shows up on my profile, which makes me very happy.

]]> By: peirz http://meshula.net/wordpress/?p=174&cpage=1#comment-6451 peirz Thu, 26 Jun 2008 14:27:55 +0000 http://meshula.net/wordpress/?p=174#comment-6451 Once the binding is done, the map will never change, so I'd suggest a sorted vector instead. It's not a premature optimization, rather it's simplifying the code by making it explicit that once you've got things set up, it's just a lookup from then on. Although, [] is more readable than lower_bound and all that. It also gets rid of the redundancy where the name is stored in the attribute _and_ in the map as a key. Once the binding is done, the map will never change, so I’d suggest a sorted vector instead. It’s not a premature optimization, rather it’s simplifying the code by making it explicit that once you’ve got things set up, it’s just a lookup from then on. Although, [] is more readable than lower_bound and all that. It also gets rid of the redundancy where the name is stored in the attribute _and_ in the map as a key.

]]>