On Thursday, June 7, 2012 1:53:30 PM UTC-4, Tom Hume wrote:
Hi
I'm trying to generate a sequence which corresponds to a breadth-first
search of a very wide, deep tree... and I'm hitting memory problems when I
go too far along the sequence. Having asked around on the IRC channel and
(dorun (take 2000 (add-layer)))
take holds on to the head, because that is what it returns. Try changing
take to drop.
Take is lazy though, and dorun should drop the head, so that shouldn't be a
problem.
The problem here is not an holding onto the head issue. Lots of memory is
being
Thanks David. Could you recommend any other types of search, or a different
algorithm entirely, which might give me the same output in a sequence?
On Fri, Jun 8, 2012 at 1:29 PM, David Powell djpow...@djpowell.net wrote:
(dorun (take 2000 (add-layer)))
take holds on to the head, because
Nope, that doesn't make any difference; if something's keeping hold of the
head, it isn't the take:
= (first (drop 1999 (add-ch)))
OutOfMemoryError Java heap space clojure.lang.RT.cons (RT.java:552)
= (first (drop 1999 (add-layer)))
OutOfMemoryError Java heap space
Are you able to go to billions of items with the first version? I'm going
to be using this to work with truly vast sequences, so if there's any sort
of memory issue with the approach, I would expect it to be exposed at some
point - given that I can only throw finite amounts of memory at this.
Hi
I'm trying to generate a sequence which corresponds to a breadth-first
search of a very wide, deep tree... and I'm hitting memory problems when I
go too far along the sequence. Having asked around on the IRC channel and
looked here, the number 1 cause of such problems is inadvertently
The first version completes without problems.
Make sure your JVM has plenty of memory available to garbage collect.
Look at the -Xmx setting (for instance with leiningen :jvm-opts [-
Xmx2g -server])
REPL started; server listening on localhost port 18885
user= (def atoms '(a b c))
#'user/atoms
