Hello all, I am in the process of plugging Chris' memory tracking tool (watch.sh) into the daily smoke test. There seem to be some problems with the current build but I did get some results using the 11/21 build. A graph of the results is attached. For this test I opened about 120 url's sequentially and immediate opened the same set of urls a second time. Notice the big jump in memory (15+ mb) about halfway through the graph and again just before the end. This these are results of opening one of the two urls http://wam.umd.edu/~soref/stress.html or www.logitech.com which I intentionally added because they are reputed to be memory hogs. Interestingly, not only is the offending url a memory hog but the browser never recovers from the memory jump. This looks pretty bad to me. Notice also that the memory consumption never actually levels off as I had thought it might be doing when I ran a single pass through the urls. Although the rate of growth in consumption is slow, over 250 urls it is clear that it is steady. One other observation: Comparing the two halves of the graph confirms that the memory is pattern is virtually identical for the same set of urls. I thought it useful to try to confirm that results from one pass would be duplicated in subsequent passes. That appears to be the case from this one run at least. Thanks, Curt
