Status of all platform builds of jdk7u-osx repos

Kelly O'Hair kelly.ohair at oracle.com
Mon Dec 5 09:59:26 PST 2011


I can wait, not like I don't have a dozen other things to do, like everyone else. ;^)

I may integrate some make/jprt.properties file changes, which are 100% harmless to everyone
but JPRT. I want to set it up so that all JPRT jobs add macosx builds and tests by default.

On the printf issue,  I failed to see any man page or documentation on any system that implied
that printf even took options, much less accepting the -- convention.
At first I thought we just needed to add a -- argument, but I wonder if that is the right thing to do
if there is no documentation that -- is officially accepted. :^(
I was thinking we just need to change the -- to @@ or maybe a smiley face :^)

-kto

On Dec 5, 2011, at 9:22 AM, Michael McMahon wrote:

> On 05/12/11 16:27, Kelly O'Hair wrote:
>> OK. No problem.
>> 
>> Does someone have a fix for the 'printf --' Mac issue? Or should I file a CR and take care of it?
>> 
> I have a fix, and I was going to make a few miscellaneous build changes under one CR
> in the next day or so. If you need a fix sooner then this seems to work, but haven't tested
> it on Windows yet. The initial "-" in the printf below is interpreted as a format char on
> macos.
> 
> diff -r 870fd5101f66 make/Defs-internal.gmk
> --- a/make/Defs-internal.gmk    Wed Nov 16 16:09:43 2011 -0800
> +++ b/make/Defs-internal.gmk    Mon Dec 05 09:18:18 2011 -0800
> @@ -79,7 +79,7 @@
> # Find all build_time_* files and print their contents in a list sorted
> # on the name of the sub repository.
> define ReportBuildTimes
> -$(PRINTF) "-- Build times ----------\nTarget %s\nStart %s\nEnd   %s\n%s\n%s\n-------------------------\n" \
> +$(PRINTF) "\55- Build times ----------\nTarget %s\nStart %s\nEnd   %s\n%s\n%s\n-------------------------\n" \
> $1 \
> "`$(CAT) $(BUILDTIMESDIR)/build_time_start_TOTAL_human_readable`" \
> "`$(CAT) $(BUILDTIMESDIR)/build_time_end_TOTAL_human_readable`" \
> 
> 
>> -kto
>> 
>> On Dec 5, 2011, at 8:12 AM, Michael McMahon wrote:
>> 
>>> The interface was there (in hotspot) but I presume due an oversight wasn't exported.
>>> I guess that's beside the point though. From a process point of view I probably should
>>> have disabled that functionality to avoid the dependency.  So, mea culpa.
>>> 
>>> I figured that since the forest wasn't really usable up to this point, it wouldn't
>>> cause a problem for anyone. But, I agree it might be better to avoid doing this in future
>>> 
>>> - Michael.
>>> 
>>> On 05/12/11 16:00, Kelly O'Hair wrote:
>>>> So how did the jdk changes get integrated without the hotspot changes?
>>>> That seems like a process problem. You should NEVER add a dependency on an interface that doesn't exist yet.
>>>> 
>>>> Was this a jdk7u-osx issue or a jdk7u-dev issue?
>>>> 
>>>> -kto
>>>> 
>>>> On Dec 4, 2011, at 1:15 PM, Paul Hohensee wrote:
>>>> 
>>>>> Yes, we're missing something in Hotspot.  The fix is in hs23 b07, which is
>>>>> being PIT'ed now and will be pushed up to jdk7u-dev Tuesday, if all goes
>>>>> well.  We can pull it down to jdk7u-osx as soon as it's pushed to jdk7u-dev.
>>>>> 
>>>>> Paul
>>>>> 
>>>>> On 12/4/11 3:44 PM, Kelly O'Hair wrote:
>>>>>> FYI...
>>>>>> 
>>>>>> I ran the jdk7u/jdk7u-osx forest through our internal build system that builds all the platforms
>>>>>> plus the Mac.
>>>>>> 
>>>>>> Only windows builds completed successfully.
>>>>>> 
>>>>>> On Solaris and Linux, there is some kind of issue with a missing JVM external:
>>>>>>     JVM_SetNativeThreadName
>>>>>> when building libjava.so. Are we missing something in hotspot?
>>>>>> 
>>>>>> On the Mac, the builds appeared to complete then die at the last minute with a complaint about
>>>>>> options passed into printf. It appears that the Mac does not like the "--" characters being passed
>>>>>> in (file make/Defs-internal.gmk in define ReportBuildTimes).
>>>>>> 
>>>>>> ########################################################################
>>>>>> ##### Leaving jdk for target(s) sanity all docs images             #####
>>>>>> ########################################################################
>>>>>> ##### Build time 00:14:15 jdk for target(s) sanity all docs images #####
>>>>>> ########################################################################
>>>>>> 
>>>>>> /usr/bin/printf: illegal option -- -
>>>>>> usage: printf format [arguments ...]
>>>>>> make[1]: *** [generic_build_repo_series] Error 1
>>>>>> make: *** [build_product_image] Error 2
>>>>>> 
>>>>>> 
>>>>>> -kto
>>>>>> 
> 



More information about the macosx-port-dev mailing list