Bug 38889 – [chromium] don't build DRT in chromium checkouts due to a circular gyp dependency

• Status: RESOLVED
• Resolution: FIXED
• Component: New Bugs

[Tony Chang]

[chromium] don't build DRT in chromium checkouts due to a circular gyp dependency

[Tony Chang]

Created attachment 55670 [details]
Patch

[Tony Chang]

This is unfortunate, but there's currently a circular dependency if you try to build DRT in a chromium checkout.  Disabling for now until we can work out a gyp refactoring to avoid this.

[Kent Tamura]

What circulation occurs for now?

[Tony Chang]

(In reply to comment #3)
> What circulation occurs for now?

webkit.gyp:pull_in_webkit_unit_tests -> WebKit.gyp -> webkit.gyp:webkit_support.

I'm not sure why we need pull_in_webkit_unit_tests, but I figure I should fix the build before trying to fix the circular dependency.

[Tony Chang]

(In reply to comment #4)
> (In reply to comment #3)
> > What circulation occurs for now?
> 
> webkit.gyp:pull_in_webkit_unit_tests -> WebKit.gyp -> webkit.gyp:webkit_support.
> 
> I'm not sure why we need pull_in_webkit_unit_tests, but I figure I should fix the build before trying to fix the circular dependency.

According to the changelog, this is needed to add webkit_unit_tests to webkit.sln and chrome.sln.  Hmm, maybe this can be windows only since the circular gyp check only happens on mac.

[Kent Tamura]

Comment on attachment 55670 [details]
Patch

Ok, I understand.

[Tony Chang]

Committed r59145: <http://trac.webkit.org/changeset/59145>

[Tony Chang]

There's also a dependency from test_shell.gypi to WebKit.gyp, but I think that can be solved by moving test_shell.gypi to test_shell.gyp and having a pull_in_test_shell target for windows.

[Darin Fisher (:fishd, Google)]

> I'm not sure why we need pull_in_webkit_unit_tests, but I figure I should fix the build before trying to fix the circular dependency.

In chromium, trunk/src/webkit/webkit.gyp should pull in everything related to webkit so that if you are just doing webkit development, you can just use that.  I don't see why that should be Windows only.

Maybe we can leverage the inside_chromium_build variable to help resolve some of these issues.