Code to load the software renderer and initialize it. Build scripts copy the ...

Code to load the software renderer and initialize it. Code to force the software renderer enabled with registry entry even if hardware is capable. Build scripts copy the software renderer dll to the plugin directory on an intall build.


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=19070

[vangelis, 2009-06-23 20:49:37]

http://codereview.chromium.org/147039/diff/1/2
File core/win/d3d9/install_check.cc (right):

http://codereview.chromium.org/147039/diff/1/2#newcode219
Line 219: *error = "Failed to compute new plugin location.";
Change 'new plugin location' to 'software renderer location'  ?

http://codereview.chromium.org/147039/diff/1/4
File core/win/d3d9/renderer_d3d9.cc (right):

http://codereview.chromium.org/147039/diff/1/4#newcode363
Line 363: if (SUCCEEDED(RegCreateKeyEx(HKEY_CURRENT_USER,
Do we want to do RegCreateKeyEx or RegOpenKeyEx? Is there a reason we should be
creating that key?

http://codereview.chromium.org/147039/diff/1/4#newcode382
Line 382: return true;
should be:
enabled = true

instead of:
return true

so that you get to call RegCloseKey() on it.

http://codereview.chromium.org/147039/diff/1/4#newcode428
Line 428: // Report the hardware error. An error with the software renderer
should
Not 'hardware error' at this point.

[vangelis, 2009-06-23 21:00:08]

BTW could we make a unit test for the s/w renderer?  It seems to me that we
should add an API to the client that indicates whether you're running in s/w or
not.

[apatrick, 2009-06-23 21:23:38]

http://codereview.chromium.org/147039/diff/1/2
File core/win/d3d9/install_check.cc (right):

http://codereview.chromium.org/147039/diff/1/2#newcode219
Line 219: *error = "Failed to compute new plugin location.";
On 2009/06/23 20:49:37, vangelis wrote:
> Change 'new plugin location' to 'software renderer location'  ? 

Done.

http://codereview.chromium.org/147039/diff/1/4
File core/win/d3d9/renderer_d3d9.cc (right):

http://codereview.chromium.org/147039/diff/1/4#newcode363
Line 363: if (SUCCEEDED(RegCreateKeyEx(HKEY_CURRENT_USER,
On 2009/06/23 20:49:37, vangelis wrote:
> Do we want to do RegCreateKeyEx or RegOpenKeyEx? Is there a reason we should
be
> creating that key?

Done.

http://codereview.chromium.org/147039/diff/1/4#newcode382
Line 382: return true;
On 2009/06/23 20:49:37, vangelis wrote:
> should be:
> enabled = true
> 
> instead of:
> return true
> 
> so that you get to call RegCloseKey() on it.

Done.

http://codereview.chromium.org/147039/diff/1/4#newcode428
Line 428: // Report the hardware error. An error with the software renderer
should
On 2009/06/23 20:49:37, vangelis wrote:
> Not 'hardware error' at this point.

It is here. Check out line 419.

[vangelis, 2009-06-23 22:11:20]

LGTM