[webkit-reviews] review denied: [Bug 42612] DRT should allow test scripts to access WebCore internal states for setup/verification : [Attachment 62805] rebaselined

[bugzilla-daemon at webkit.org]

Adam Barth <abarth at webkit.org> has denied MORITA Hajime <morrita at google.com>'s
request for review:
Bug 42612: DRT should allow test scripts to access WebCore internal states for
setup/verification
https://bugs.webkit.org/show_bug.cgi?id=42612

Attachment 62805: rebaselined
https://bugs.webkit.org/attachment.cgi?id=62805&action=review

------- Additional Comments from Adam Barth <abarth at webkit.org>
WebCore/Configurations/FeatureDefines.xcconfig:47
 +  ENABLE_CORE_LAYOUT_TEST_CONTROLLER_Production = ;
I'm not sure this is a good approach.  Differences between Production and
Release are bad because it means we're not testing the same thing we're
shipping.  Also, it seems very likely that some port or other will get confused
and ship WebCore::LTC to users.  I'm not sure what the consequences are today,
but as it grows, I'm sure it will eventually be disaster.

Also, as a consequence, we won't be able to run these tests on production
builds.  That means, we'll lose out on test coverage of the actual bits we
ship.  There's got to be a better approach.

Here's a strawman proposal:

1) WebCore exports a C API for testing that consumed by a cross-platform layout
test controller object that's build in a different project (DRT).

In that approach, we save the per-port hassle of adding features to LTC and we
don't risk shipping those API to the web.