Bug 37183
| Summary: | new-run-webkit-tests should ensure font smoothing is in a good state for chromium-mac | ||
|---|---|---|---|
| Product: | WebKit | Reporter: | Dirk Pranke <dpranke> |
| Component: | Tools / Tests | Assignee: | Nobody <webkit-unassigned> |
| Status: | RESOLVED FIXED | ||
| Severity: | Normal | CC: | dcheng, hbono, thakis |
| Priority: | P2 | ||
| Version: | 528+ (Nightly build) | ||
| Hardware: | PC | ||
| OS: | OS X 10.5 | ||
Dirk Pranke
A number of the image tests may fail if the user's computer is not using the default font smoothing settings on OS X. I believe run-webkit-tests gets around this by explicitly changing the font-smoothing settings, but new-run-webkit-tests doesn't do this.
The correct setting is probably "defaults -currentHost write -g AppleFontSmoothing -int 0" (or maybe -int 1)
| Attachments | ||
|---|---|---|
| Add attachment proposed patch, testcase, etc. |
Mark Rowe (bdash)
DumpRenderTree sets the user default in its process, not run-webkit-tests. Setting it globally would be a bad idea.
Dirk Pranke
(In reply to comment #1)
> DumpRenderTree sets the user default in its process, not run-webkit-tests.
> Setting it globally would be a bad idea.
Good point. I had just gotten far enough in looking through run-webkit-tests to see that run-webkit-tests wasn't doing it :)
-- Dirk
Hironori Bono
(In reply to comment #2)
> (In reply to comment #1)
> > DumpRenderTree sets the user default in its process, not run-webkit-tests.
> > Setting it globally would be a bad idea.
>
> Good point. I had just gotten far enough in looking through run-webkit-tests to
> see that run-webkit-tests wasn't doing it :)
Our test_shell also sets the user default in its process when the |classic_theme| value is true. (On Mac, its default value was true.) As far as I investigated, it seems (probably) your r26203 (*1) changed its default value to false and prevented setting the user default on Mac. So, I would think the easiest solution is setting its default value back to true on Mac.
(*1) <http://src.chromium.org/viewvc/chrome?view=rev&revision=26203>
Regards,
Hironori Bono
Nico Weber
The upstream bug of this, http://code.google.com/p/chromium/issues/detail?id=42352, has been fixed. So this can probably be closed as well?
Dirk Pranke
Yes, I think it's safe to close this, at least as far as TestShell goes (and I think the DRT port has this fixed as well).