Bug 91752 – [chromium][Mac] Switch the MACOSX_DEPLOYMENT_TARGET to 10.6

• Component: WebKit Misc.
• Status: RESOLVED FIXED

[Robert Sesek]

Downstream Chromium has just changed its build SDK from 10.5 to 10.6, which required no changes. However changing the deployment target from 10.5 to 10.6 generates some -Wdeprecated-declaration warnings that are converted to errors. This bug will track fixing those.

[Robert Sesek]

Created attachment 153282 [details]
ATSUI,v1

[Robert Sesek]

Created attachment 153287 [details]
ColorSpaceWarnings,v1

[Robert Sesek]

Jeremy: Please review ATSUI,v1 patch to disable ATSUI support on Chromium WebKit. This should not be needed now that downstream does not support 10.5.

Mark: Please review ColorSpaceWarnings,v1 to disable #pragmas for a deprecated symbol for which there is no replacement.

[Eric Seidel (no email)]

Comment on attachment 153282 [details]
ATSUI,v1

Great!  Won't we need to update our pixel results?

[Eric Seidel (no email)]

You should probably CC the current gardener.

[Eric Seidel (no email)]

Comment on attachment 153287 [details]
ColorSpaceWarnings,v1

What does the Mac port do?  Why don't they run into this?

[Robert Sesek]

(In reply to comment #6)
> (From update of attachment 153287 [details])
> What does the Mac port do?  Why don't they run into this?

As far as I could find, they don't have an equivalent to WebKit::WebScreenInfo.isMonochrome implemented on Mac.

[Robert Sesek]

(In reply to comment #4)
> (From update of attachment 153282 [details])
> Great!  Won't we need to update our pixel results?

I don't *think* so because only 10.5 should have been using the ATSUI path, and we shouldn't have any testers on 10.5 anymore. I think 10.6 and above would all be using CoreText for font layout. I tried running locally, but I'm on 10.8 now so it's hard to tell if any of the pixel tests are wrong because they changed the subpixel rendering in this version.

[WebKit Review Bot]

Comment on attachment 153282 [details]
ATSUI,v1

Clearing flags on attachment: 153282

Committed r123161: <http://trac.webkit.org/changeset/123161>

[Eric Seidel (no email)]

Comment on attachment 153287 [details]
ColorSpaceWarnings,v1

It seems CGGetColorSpaceModel is your replacement: https://developer.apple.com/library/mac/#documentation/graphicsimaging/reference/CGColorSpace/Reference/reference.html

At least that's what a quick search lead me to.

[Eric Seidel (no email)]

CGColorSpaceGetModel, rather.

[Eric Seidel (no email)]

You'll have to upload again, your attachment was lost during the window with the failed migration.

[Robert Sesek]

Created attachment 153509 [details]
ColorSpaceWarnings,v2

[Eric Seidel (no email)]

Comment on attachment 153509 [details]
ColorSpaceWarnings,v2

Do we need to wrap this all in some sort of exception safely? since this is obj-c in c++?  I know we do similar things in WebKit Mac.

[Robert Sesek]

No, this doesn't throw ObjC exceptions, which also are different than C++ exceptions.

[WebKit Review Bot]

Comment on attachment 153509 [details]
ColorSpaceWarnings,v2

Clearing flags on attachment: 153509

Committed r123256: <http://trac.webkit.org/changeset/123256>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.

[Jeremy Moskovich]

Robert: Thanks for looking into this!!

Do you want to file another bug for the ATSUI removal or reopen this one?  Are there any other ports that still use ATSUI?  (Looks like there aren't, but we probably want to check with Mitz).

If not then it'd probably be good to just get rid of all the code and the flag entirely.

[Jeremy Moskovich]

For future reference, mitz obsolted ATSUI in r123181 .