Instead of "reusing" the AppleWin port when running tests on WinCairo, we should add it as a port, and add baseline results for it.
Created attachment 264212 [details] Patch
Comment on attachment 264212 [details] Patch I would suggest trying to share as much "Apple Windows" and "WinCairo" results as possible, though perhaps in practice there are too many differences in font handling, antialiasing, etc.
(In reply to comment #2) > Comment on attachment 264212 [details] > Patch > Thanks for reviewing :) > I would suggest trying to share as much "Apple Windows" and "WinCairo" > results as possible, though perhaps in practice there are too many > differences in font handling, antialiasing, etc. Thanks, that makes sense :) I assume it means we have to change the fallback order? Currently, on Windows 10, I get the fallback order "wincairo-win10 -> wincairo -> mac -> generic".
(In reply to comment #3) > (In reply to comment #2) > > Comment on attachment 264212 [details] > > Patch > > > > Thanks for reviewing :) > > > I would suggest trying to share as much "Apple Windows" and "WinCairo" > > results as possible, though perhaps in practice there are too many > > differences in font handling, antialiasing, etc. > > Thanks, that makes sense :) I assume it means we have to change the fallback > order? > Currently, on Windows 10, I get the fallback order "wincairo-win10 -> > wincairo -> mac -> generic". Yes -- I think we have fewer tests that are compatible with 'mac' now, since the font changes made in Yosemite mean even more things differ. So in practice, the 'mac' fallback may not be that useful anymore.
Ok, thanks :) Btw, are you using the Times font from a Mac computer when you are running the layout tests on AppleWin? I ran some layout tests on AppleWin the other day, and I did not get the expected font metrics for the Times 16 font.
Committed r191678: <http://trac.webkit.org/changeset/191678>.