A couple people (Nathan and Britton) have mentioned memory problems with time series data because the StaticOutput objects sometimes hold on to hierarchies. I've taken a pass at preventing this from happening by keeping the static outputs from being instantiated until requested, and allowing the references to be passed rather than retained.
Anyway, it works for me and I've checked that the references go to zero. If somebody else that was having memory issues could take a gander that'd be great. Here's my sample script; if you run before&after the numbers it prints out at the end should be "something" and "0".
---------- Forwarded message ---------- From: Matthew Turk firstname.lastname@example.org Date: Sun, Apr 22, 2012 at 10:52 AM Subject: [yt_analysis/yt] Change time series objects to *not* instantiate pfs by default (pull request #143) To: email@example.com
A new pull request has been opened by Matthew Turk.
MatthewTurk/yt has changes to be pulled into yt_analysis/yt.
Title: Change time series objects to *not* instantiate pfs by default
This change, as described in the commit message, will prevent time series data from initially instantiating any parameter file objects. By keeping the references to pfs temporary, this will prevent hierarchies from sticking around during the course of analysis, which should improve overall memory usage.
Changes to be pulled:
-- This is an issue notification from bitbucket.org. You are receiving this either because you are the participating in a pull request, or you are following it.