Bug 36687

Summary: change new-run-webkit-tests to not use more than 4 threads by default on the Mac port
Product: WebKit Reporter: Dirk Pranke <dpranke>
Component: Tools / TestsAssignee: Dirk Pranke <dpranke>
Status: RESOLVED FIXED    
Severity: Normal CC: cjerdonek, eric, mmoss
Priority: P2    
Version: 528+ (Nightly build)   
Hardware: PC   
OS: OS X 10.5   
Attachments:
Description Flags
Patch none

Description Dirk Pranke 2010-03-26 18:25:58 PDT
new-run-webkit-tests seems to be unstable w/ more than 4 threads running in parallel (see bug 36622). Change the default until that bug is fixed.
Comment 1 Dirk Pranke 2010-03-26 18:27:38 PDT
Created attachment 51803 [details]
Patch
Comment 2 Chris Jerdonek 2010-03-27 12:44:08 PDT
How does this issue relate to this code?

http://trac.webkit.org/browser/trunk/WebKitTools/Scripts/webkitpy/common/system/executive.py#L121
Comment 3 Dirk Pranke 2010-03-28 17:21:41 PDT
(In reply to comment #2)
> How does this issue relate to this code?
> 
> http://trac.webkit.org/browser/trunk/WebKitTools/Scripts/webkitpy/common/system/executive.py#L121

new-run-webkit-tests doesn't use that code, and I have no idea where that code is used. I'm not sure if that answers your questin?
Comment 4 Chris Jerdonek 2010-03-28 17:32:29 PDT
(In reply to comment #3)
> (In reply to comment #2)
> > How does this issue relate to this code?
> > 
> > http://trac.webkit.org/browser/trunk/WebKitTools/Scripts/webkitpy/common/system/executive.py#L121
> 
> new-run-webkit-tests doesn't use that code, and I have no idea where that code
> is used. I'm not sure if that answers your questin?

Superficially, it looks like we have two ways of getting the CPU (core?) count in our code base.  If they are doing the same thing, we might want to combine those code paths.
Comment 5 Dirk Pranke 2010-03-29 10:47:09 PDT
(In reply to comment #4)
> (In reply to comment #3)
> > (In reply to comment #2)
> > > How does this issue relate to this code?
> > > 
> > > http://trac.webkit.org/browser/trunk/WebKitTools/Scripts/webkitpy/common/system/executive.py#L121
> > 
> > new-run-webkit-tests doesn't use that code, and I have no idea where that code
> > is used. I'm not sure if that answers your questin?
> 
> Superficially, it looks like we have two ways of getting the CPU (core?) count
> in our code base.  If they are doing the same thing, we might want to combine
> those code paths.

Yes, you are correct. There's a lot of duplication between the port/* interface and code written outside of the layout_tests directory, and we haven't really tried to reduce that yet. Nobody has yet proposed what the right structure is, and I've been mostly focused on simply getting the new-run-webkit-tests to be bug-free and feature-equivalent to the Perl code.
Comment 6 Dimitri Glazkov (Google) 2010-03-29 13:15:07 PDT
Comment on attachment 51803 [details]
Patch

ok.
Comment 7 Dirk Pranke 2010-03-29 15:35:43 PDT
Comment on attachment 51803 [details]
Patch

Clearing flags on attachment: 51803

Committed r56746: <http://trac.webkit.org/changeset/56746>
Comment 8 Dirk Pranke 2010-03-29 15:35:48 PDT
All reviewed patches have been landed.  Closing bug.