> I'm not sure which abbreviations mechanism you mean; if 
> you're talking 
> about BTreeStringKeys, don't go there - it's way too messy.  What you 
> want to do should be at the node level anyway, and not worry 
> about the 
> details of how the keys are stored.
OK forgotten :P

> 
> And yes, my BTree class could definitely use some extra iteration 
> protocol for ranges of keys and so on.  Maybe I'll find some 
> time today 
> to release a new version.
I'm happy to know that.

> 
> However, for what you're asking, my guess is that the best data 
> structure is a Trie.  I was talking to someone recently about 
> this who 
> was going to implement one; hopefully he'll pipe up and you 
> can compare 
> notes with him.
I was reading about tries. I think the tries are pretty much like we
need but the implementation is not trivial enough to develop it now.
Perhaps at the long run.
Anyway I think the IndexedCollection could help a lot by now. 

Please let me know I you has the range key selection of the BTree.

Regards,

Sebastián Sastre
ssastre at seaswork.com.ar
www.seaswork.com.ar

---
Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.760 / Virus Database: 509 - Release Date: 10/09/2004