It lists --child-processes as a valid commandline argument, but it always creates 1 process. I think this is a regression. It's fine to default to 1, but if you pass the flag, it should respect it. I had noticed recently the run-webkit-tests was taking forever even when I'd run with multiple child processes. I expect I might not be the only one surprised by this.
That would be an odd regression. I'll look on monday, now that I'm done with the CSS regression stuff. If you run "new-run-webkit-tests" directly does child-processes work as you expect? the "run-webkit-tests" wrapper overrides --child-process to 1.
It works correctly if I just new-run-webkit-tests directly.
OK. Should be easy to teach the wrapper to avoid overriding you. But we could also mark this wontfix. I'm working on NRWT again as of tomorrow. Done with CSS regression distractions.
Created attachment 111151 [details] Patch
Sorry if this perl is crap. I bet there's a simple one-liner for this. I'm just a perl noob.
Comment on attachment 111151 [details] Patch Check out usingWebKit2() and usingLeaks() for some ideas about how to do this more compactly.
Created attachment 111153 [details] Patch
Comment on attachment 111153 [details] Patch Clearing flags on attachment: 111153 Committed r97567: <http://trac.webkit.org/changeset/97567>
All reviewed patches have been landed. Closing bug.