On 2015-02-04 22:16, Phil Race wrote:
Was there an implicit nod that the 8 fix was OK if I updated it as below,
or was my reply too buried in the email for anyone to notice the update.
Sorry, I missed it. Looks good. Thank you for sorting out this mess!
/Magnus
The updated 8 fix is (as previously noted) here :-
http://cr.openjdk.java.net/~prr/8072116.8u.1/
and passed JPRT.
-phil.
On 02/03/2015 08:56 AM, Phil Race wrote:
On 02/03/2015 05:36 AM, Magnus Ihse Bursie wrote:
On 2015-02-02 19:33, Phil Race wrote:
http://bugs.openjdk.java.net/browse/JDK-8071710 added a depedency
check but it was not working because this depends on file
declaration order.
This resolves that which can cause occasional Solaris build
failures by
ensuring that HEADLESS is declared before FONTMANAGER.
JDK 9 fix : http://cr.openjdk.java.net/~prr/8072116/
JDK8u fix : http://cr.openjdk.java.net/~prr/8072116.8u
FWIW I've not seen it fail on 8u but it seems like it must be
possible there too ..
I see now that the fix for 8u in the original JDK-8071710 did not
get the same dependency declaration as did the 9 fix. It still says:
948 ifneq (, $(findstring $(OPENJDK_TARGET_OS), solaris aix))
949 $(BUILD_LIBFONTMANAGER): $(BUILD_LIBAWT_XAWT)
950 endif
instead of depending on libawt_headless.
Strange: I had that in the previous webrev ..
http://cr.openjdk.java.net/~prr/8071710.8u.1/make/lib/Awt2dLibraries.gmk.sdiff.html
.. must have been lost.
Rectifying that here :- http://cr.openjdk.java.net/~prr/8072116.8u.1/
Still, that doesn't explain why 8 built reliably for me since there was
no HEADLESS dependency ..
-phil.
Apart from this, the fix looks good.
/Magnus
