Pull request: https://github.com/aibasel/downward/pull/101

Merged.

Thanks! Feel free to merge.

Good point. I adapted the changelog message.

Looks good to me with one comment: I see in the diff that some functionality is not available on Windows. Mention this in the changelog? Not necessary if this is just moved code rather than new code.

Makes sense. In my pull request (see above), I now print the overall planner runtime. Printing peak memory is not as straightforward. Can you have a look at the code, please?

Makes sense to me, but can we combine this with also printing total planner runtime in the same way? To me how much time was actually taken is more basic information than how much time was allowed.

It would be nice to also print the overall peak memory usage, but this would be tougher with the current code, whereas overall time usage should be easy enough, right? I assume the driver already needs to know this in order to set the timeouts for the processes it calls. If we can also make overall peak memory usage work, even better.

In order to compute scores for planner_time and planner_memory with Downward Lab, I propose to let the driver output the overall time and memory limits. The limits are needed as upper bounds for computing the logarithmic scores. Here is the proposed output format:


INFO     planner time limit: 22s
INFO     planner memory limit: 100 MB

INFO     Running translator.
INFO     translator stdin: None
INFO     translator time limit: 21.96s
INFO     translator memory limit: 100 MB
...


Thoughts?