RFR: JDK-8071781: Bootcycle build fails on macosx

[Magnus Ihse Bursie]

On 2015-01-28 14:58, Erik Joelsson wrote:
> Hello,
>
> Please review this small fix for bootcycle-images on macosx. The space 
> in the patsubst call caused HOTSPOT_DIST to contain a leading space 
> which in turn made hotspot/make/bsd/makefiles/universal.gmk fail to 
> run lipo on the files in HOTSPOT_DIST in the second phase bootcycle 
> build.
>
> Bug: https://bugs.openjdk.java.net/browse/JDK-8071781
> Patch inline:
> diff --git a/common/autoconf/bootcycle-spec.gmk.in 
> b/common/autoconf/bootcycle-spec.gmk.in
> --- a/common/autoconf/bootcycle-spec.gmk.in
> +++ b/common/autoconf/bootcycle-spec.gmk.in
> @@ -48,8 +48,9 @@
>  # The bootcycle build has a different output directory
>  OLD_BUILD_OUTPUT:=@BUILD_OUTPUT@
>  BUILD_OUTPUT:=$(OLD_BUILD_OUTPUT)/bootcycle-build
> -# The HOTSPOT_DIST dir is not defined relative to BUILD_OUTPUT in 
> spec.gmk
> -HOTSPOT_DIST:=$(patsubst $(OLD_BUILD_OUTPUT)%, $(BUILD_OUTPUT)%, 
> $(HOTSPOT_DIST))
> +# The HOTSPOT_DIST dir is not defined relative to BUILD_OUTPUT in 
> spec.gmk. Must not
> +# use space in this patsubst to avoid leading space in HOTSPOT_DIST.
> +HOTSPOT_DIST:=$(patsubst 
> $(OLD_BUILD_OUTPUT)%,$(BUILD_OUTPUT)%,$(HOTSPOT_DIST))
>  SJAVAC_SERVER_DIR:=$(patsubst $(OLD_BUILD_OUTPUT)%, $(BUILD_OUTPUT)%, 
> $(SJAVAC_SERVER_DIR))
>
>  JAVA_CMD:=$(BOOT_JDK)/bin/java
>

Looks good to me. But oh how a single small space can cause the entire 
build to come into a flaming crash. :-)

/Magnus