It wouldn't change the performance profile of existing programs because 
currently there isn't a way to concat two vectors and get a new vector. I don't 
envision changing the behavior of clojure.core/concat, that should still return 
a seq. It seems better to create a new function like 'joinvec' that is a more 
performant alternative to concat. This is similar to subvec, which is a more 
performant alternative to take and drop.

Another question is whether to change the behavior of subvec, which currently 
holds onto the entire vector, or create a new method like 'splitvec' to provide 
the new RRB-based splitting.



On Nov 15, 2011, at 8:59 AM, Karl Krukow <karl.kru...@gmail.com> wrote:

> 
> 
> On Tuesday, November 15, 2011, Justin Balthrop <jus...@justinbalthrop.com> 
> wrote:
> 
> > My understanding from the talk was that RRB-Trees have performance
> > identical to PersistentVector as long as you don't concat or split
> > them. So why not just replace the PersistentVector implementation with
> > an RRB-Tree?
> 
> I guess that could be a possibility but that would change the performance 
> profile of some programs that actually use concat,  ie concat would speed up 
> and the other operations would slow (was it about 60%? can't remember the 
> numbers here). So I think it should be opt in.
> 
> Karl
> -- 
> You received this message because you are subscribed to the Google
> Groups "Clojure" group.
> To post to this group, send email to clojure@googlegroups.com
> Note that posts from new members are moderated - please be patient with your 
> first post.
> To unsubscribe from this group, send email to
> clojure+unsubscr...@googlegroups.com
> For more options, visit this group at
> http://groups.google.com/group/clojure?hl=en

-- 
You received this message because you are subscribed to the Google
Groups "Clojure" group.
To post to this group, send email to clojure@googlegroups.com
Note that posts from new members are moderated - please be patient with your 
first post.
To unsubscribe from this group, send email to
clojure+unsubscr...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/clojure?hl=en

Reply via email to