On Sun, 16 Feb 2014 21:58:08 -0500, Sean Luke <[log in to unmask]> wrote:
>On Feb 16, 2014, at 9:50 PM, Raymond Shpeley <[log in to unmask]>
>> Well, I meant it didn't show an early termination due to an error... I should
>> verbosity on, shouldn't I?
>What error would you think would cause ECJ to terminate early?
I didn't expect one would actually. Why terminate a run if there's an issue with
reporting? I was pointing out the obvious with the idea in mind that someone
else with the same or similar problem might benefit from the comment. But, I
really should have said more about it such as...
This is rather obvious, but the run was not terminated early due to an error so
it points more toward either a parameter setup issue or some other
incompatibility with stat.gather-full=true.
I've gone over my results again. The reason why the elitism part of Tutorial 4
gave me an error message is because of,
-p stat=ec.gp.koza.KozaShortStatistics -p stat.gather-full=true
for the first run of the tutorial.
This explains why I didn't get an error in the first run and why I did for the
elitism run. The package for stat.gather-full was only loaded for the elitism run.
In adding either,
and by adding,
-p stat.do-time=true -p stat.do-size=true
I get what looks like time and size info in a short information format. At the
moment I mostly don't know what columns are associated with what data.
However, I was expecting to get something like what I saw on the tutorial
Initialization: 0.312 secs total, 24692 nodes, 79141.02 nodes/sec
Evaluating: 0.221 secs total, 85624 nodes, 387438.9 nodes/sec
Breeding: 0.614 secs total, 60932 nodes, 99237.78 nodes/sec
Initialization: 1109.9062 KB total, 24692 nodes, 22.246923 nodes/KB
Evaluating: 1.6875 KB total, 85624 nodes, 50740.15 nodes/KB
Breeding: 1782.4219 KB total, 60932 nodes, 34.184948 nodes/KB
So, is this what was deprecated? The textual view of time and size information
in deference to the short format?