...
Query File: This is the query file you want to evaluate. It should deliver at least one executable, physical query plan. Remember that the query will be executed more than once, so you should design the script to be re-executable. This means for example, the query file may not try to install new sources which are already installed. Hint: you may use something like "#DROPALLSOURCES" or "#IFSRCDEF/#IFSRCNDEF" in Odysseus Script.
New (2017.02.16): There is now the possibility to add a Setup File and a Tear Down file. Similar to JUnit these scripts are called directly before and after the run (e.g. to register sources). These scripts can ran before and after each run or onyl once before and after the whole runs (Setup/Tear down every Run).
Folder for processing results:This is a the folder where the evaluation dumps the results. For each new run of an evaluation, a new subfolder (named by date and time) is created.
...
Measure Latency: If it is activated, the evaluation job adds the his pre-transformation handler to measure the latency of the query. The pre-transformation handler changes the query files (e.g. it activates the metadata latency by adding "#METADATA Latency" to the query. Furthermore, it adds the LatencyCalculationPipe and a FileSink to each root operator of the query. Therefore, if your query has different sinks, you will also have different latency results. TODO: This has changed. Currently your sources must contain valid metadata for processing!!
Create plots for latencies: If activated, the evaluation will use the merged files to create some plots, which are stored within the plot-folder (look above)
...
If you stop a query by hand (e.g. using the Queries view) - one evaluation run is canceled and the next run will be installed and started automatically.