Bug 111726

Summary: run-pert-tests should have a --test-runner-count option
Product: WebKit Reporter: Glenn Adams <glenn>
Component: Tools / TestsAssignee: Glenn Adams <glenn>
Severity: Normal CC: abarth, benjamin, dpranke, eric, rniwa, webkit.review.bot
Priority: P2    
Version: 528+ (Nightly build)   
Hardware: Unspecified   
OS: Unspecified   
Bug Depends on:    
Bug Blocks: 77037    
Description Flags
Patch none

Description Glenn Adams 2013-03-07 07:04:02 PST
Given the recent addition of process_run_count argument to PerfTest(), which controls the number of samples taken when performing an individual test in a test set, the value of this argument should be exposed as a command line option, e.g., --sampleCount.

Note that this is different from the --repeat argument proposed in bug 100030, which controls the number of times the entire test set is run. In particular, the number of samples (controlled by process_run_count) determine the number of data points used in statistical analysis; while different --repeat runs are not aggregated for statistical analysis in the shown results, i.e., are treated as distinct statistical result sets.
Comment 1 Ryosuke Niwa 2013-03-07 10:27:35 PST
Let's stick to the same terminology: process run count.
Comment 2 Glenn Adams 2013-03-07 14:26:52 PST
(In reply to comment #1)
> Let's stick to the same terminology: process run count.

thanks, we agreed on IRC to change this to test-runner-count (test_runner_count), for option and argument names, since this corresponds to the number of test runner (DRT/WTR) instances invoked (in serial execution - not parallel)
Comment 3 Glenn Adams 2013-03-07 19:09:59 PST
Created attachment 192128 [details]
Comment 4 WebKit Review Bot 2013-03-07 20:13:59 PST
Comment on attachment 192128 [details]

Clearing flags on attachment: 192128

Committed r145173: <http://trac.webkit.org/changeset/145173>
Comment 5 WebKit Review Bot 2013-03-07 20:14:03 PST
All reviewed patches have been landed.  Closing bug.