Switching webkit to gyp on windows.

Switching webkit to gyp on windows.


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=13480

[bradn, 2009-04-09 02:06:17]

There are a few small issues remaining (test_shell oddly builds, but not in the
chrome.sln), but this is getting close enough I thought I'd send it out for
review.
I suspect I should run this by others as well, any suggestions on who?

[bradn, 2009-04-09 02:07:55]

Adding maruel

[Nicolas Sylvain, 2009-04-09 02:22:55]

http://codereview.chromium.org/63152/diff/1/9
File chrome/test/worker/test_worker.vcproj (right):

http://codereview.chromium.org/63152/diff/1/9#newcode52
Line 52: AdditionalDependencies="urlmon.lib"
why?

[Mark Mentovai, 2009-04-09 02:59:51]

So what's the deal here, test_shell.gyp is OK but chrome.gyp needs more work? 
But the existing chrome.sln and vcprojs work with the GYP-generated webkit?

http://codereview.chromium.org/63152/diff/1/3
File webkit/webkit.gyp (right):

http://codereview.chromium.org/63152/diff/1/3#newcode135
Line 135: # Used on windows to ensure test_shell is where the builders expect it
I don't get it.  Where do builders expect it?

Do you just mean that when the builders build webkit.sln they need to build
test_shell?

[bradn, 2009-04-09 06:32:06]

http://codereview.chromium.org/63152/diff/1/9
File chrome/test/worker/test_worker.vcproj (right):

http://codereview.chromium.org/63152/diff/1/9#newcode52
Line 52: AdditionalDependencies="urlmon.lib"
On 2009/04/09 02:22:55, Nicolas Sylvain wrote:
> why?

Ooops that file was supposed to remain unchanged. Leftover.
It's been removed from the change.

[bradn, 2009-04-09 06:38:29]

So yes chrome.gyp is not currently working on windows (though I suspect it is
probably fairly close). I was basically working for chrome.exe at one point, but
I let that drift as I've been dropping things in bottom up.

This change will switch out webkit and test_shell to gyp on windows, but retain
the checked in chrome.sln and vcproj files for the chrome module. It's not my
intent to leave it in that state long, but I'm moderately afraid of doing a
larger piece. So far much small components have had little bugaboos that didn't
show up until built on the buildbots.

With this latest upload everything is now in a working state for this checkin
(simple_clipboard_impl.cc was arranged in a slightly odd way, but that's settled
now).

http://codereview.chromium.org/63152/diff/1/3
File webkit/webkit.gyp (right):

http://codereview.chromium.org/63152/diff/1/3#newcode135
Line 135: # Used on windows to ensure test_shell is where the builders expect it
On 2009/04/09 02:59:51, Mark Mentovai wrote:
> I don't get it.  Where do builders expect it?
> 
> Do you just mean that when the builders build webkit.sln they need to build
> test_shell?

On windows webkit.sln builds test_shell and active_shim. This forces that
behavior. Figure I'll sort out buildbot changes separately. (Or should we try do
it all at once?)

[bradn, 2009-04-09 06:39:30]

Adding dimitri at mark's suggestion, for webkit sanity checks.

[bradn, 2009-04-09 06:39:44]

On 2009/04/09 06:39:30, bradn wrote:
> Adding dimitri at mark's suggestion, for webkit sanity checks.
mal that is....

[bradn, 2009-04-09 07:14:57]

Spoke too soon. Passes on trybot on windows and mac. Fails some ui_tests on
linux. Any idea what might be up with these?

ui_tests
did not complete
failed 3
stdio
DoNotSniffHTMLFromImageGIF
DoNotSniffHTMLFromTextPlain
RespectNoSniffDirective

Well, I'll sort that out tomorrow.

[M-A Ruel, 2009-04-09 12:34:32]

Humm I clobbered hae141, hae137 and hae195 just in case.
ENABLE_VIDEO is down?

[bradn, 2009-04-09 21:00:26]

On 2009/04/09 12:34:32, M-A wrote:
> Humm I clobbered hae141, hae137 and hae195 just in case.
> ENABLE_VIDEO is down?

It was seems to be back in, so I've switched it back.