webrevs.5 for macosx changes to jdk7u-osx

[Alexander Strange]

On Dec 2, 2011, at 6:39 AM, Michael McMahon wrote:

> Final round of review for this changeset hopefully.
> Phil's comments from yesterday are incorporated
> and also the build.xml problem which required script files
> to be executable has been fixed.
> 
> Lastly, the changeset/webrev was generated on a forest synched with
> jdk7u (master) rather than with jdk7u-dev as I did (incorrectly)
> the last time.
> 
> Thanks,
> Michael
> 
> Changes relative to jdk7u-osx
> ----------------------------------------
> Modified files
> http://cr.openjdk.java.net/~michaelm/7113349/5/jdk7u-osx/modified/
> 
> New files
> http://cr.openjdk.java.net/~michaelm/7113349/5/jdk7u-osx/new/
> 
> Changes relative to macosx-port
> --------------------------------------------
> Modified files
> http://cr.openjdk.java.net/~michaelm/7113349/5/macosx-port/modified/

Reading the changes from macosx-port:

>  # Order is important here, trim jre after jdk image is created
> +ifeq ($(PLATFORM), macosx)
> +
>  images:: sanity-images post-sanity-images  \
>  	 $(INITIAL_IMAGE_JRE) $(EXTRA_JRE_TARGETS) $(INITIAL_IMAGE_JDK) \
>  	 trim-image-jre trim-image-jdk \
>       identify-image-jre identify-image-jdk \
>  	 process-image-jre process-image-jdk sec-files sec-files-win jgss-files \
>  	 $(EXTRA_IMAGE_TARGETS)
> +else
> +
> +images:: sanity-images post-sanity-images  \
> +	 $(INITIAL_IMAGE_JRE) $(INITIAL_IMAGE_JDK) \
> +	 trim-image-jre trim-image-jdk \
> +         identify-image-jre identify-image-jdk \
> +	 process-image-jre process-image-jdk sec-files sec-files-win jgss-files 
> +endif

Here I created $(EXTRA_IMAGE_TARGETS) to avoid having an if/else - it should just be defined as empty outside macosx.

> +# MMM: is headless option necessary?
> +ifeq ($(PLATFORM), macosx)
> +  JAVA_JVM_FLAGS = $(JAVA_HOTSPOT_DISABLE_PRINT_VMOPTIONS) -Djava.awt.headless=true
> +else
> +  JAVA_JVM_FLAGS = $(JAVA_HOTSPOT_DISABLE_PRINT_VMOPTIONS)
> +endif


IIRC not having the headless option caused build failures when building as a user without a GUI session. CToolkit might be more resilient now.

> -    mach := universal
> +# MMM: revisit when hotspot producinging universal libs
> +#   mach := universal
> +    mach := x86_64

When will that be? I thought that was already in. (also "producing")

> +ifeq ($(PLATFORM), macosx)
> +vpath %.c   $(call NativeSrcDirList,,native/com/sun/media/sound)
> +vpath %.cpp $(call NativeSrcDirList,,native/com/sun/media/sound)
> +
> +else
>  #
>  # Add to the ambient VPATH.
>  #
> -vpath %.c   $(call NativeSrcDirList,,native/com/sun/media/sound)
> -vpath %.cpp $(call NativeSrcDirList,,native/com/sun/media/sound)
> +vpath %.c $(SHARE_SRC)/native/com/sun/media/sound
> +vpath %.c $(PLATFORM_SRC)/native/com/sun/media/sound
> +vpath %.cpp $(PLATFORM_SRC)/native/com/sun/media/sound 
> +endif

This change shouldn't have any positive effect on other platforms unless NativeSrcDirList isn't working for them. Unless it's just a style change?