Let's see how the speed goes as we keep accumulating trials. If it
becomes a problem, we can re-examine pruning. But unless it becomes
a problem, let's just keep the philosophy of "keep everything".
Disk space is cheap. Hopefully, dbv3 will let us store arbitrary
amounts of data with no loss in performance. :-) :-) :-)
On Aug 28, 2007, at 11:26 AM, Ethan Mallove wrote:
> Are we not going to prune "trial" results in the new schema?
> We previously pruned trial results to improve query speed,
> but for small date range intervals this might not be
> worthwhile with the new schema. Also, I assume a "trial" MPI
> install could key to a "non-trial" test build, which would
> mean we can only prune out test runs if we decide to prune
> at all.
> mtt-devel mailing list