> On Aug 15, 2021, at 10:47 PM, David Holmes <david.hol...@oracle.com> wrote:
>
> Hi Michael,
>
> On 16/08/2021 12:15 pm, Michael Hall wrote:
>>>> It failed as before. If you are interested in the build.log it is pretty
>>>> much as above.
>>>
>>> Sorry I meant the full log, in particular the configure part that shows
>>> what tools are being used.
>>>
>>>>
>>>
>>> No invocation of a Java based tool in the build process should be setting
>>> any kind of classpath that refers to system level packages like that.
>>> However if your boot JDK is what is doing that then there may not be
>>> anything the build can do about it.
>> The config log is somewhat larger I can include it if you want or put it
>> somewhere you can see it.
>
> Somewhere I could see would be good - thanks.
http://mikehall.pairserver.com/configure.log
<http://mikehall.pairserver.com/configure.log>
>
>> I’m not quite following what you are saying. No java tool is at fault here
>> except in possibly honoring the setting of the CLASSPATH environment
>> variable.
>
> They should only honour that variable if no explicit classpath is set for the
> tool invocation, and IIUC all our build tool invocations should be setting an
> explicit classpath as you don't want the build to accidentally pick things up.
>
warning: [path] bad path element "/Applications/OpenOffice.app/Contents/lib":
no such file or directory
compiler.properties
# 0: path
compiler.warn.path.element.not.found=\
bad path element "{0}": no such file or directory
I guess I was assuming javac. I was going to try to isolate removing which
-Werror worked around things but then got Romy’s post.
You should be able to reproduce just by doing something like
export
CLASSPATH=$CLASSPATH:/Library/Frameworks/BSF4ooRexx.framework/Classes/bsf4ooRexx-v641-20190830-bin.jar
And then build on OS/X
make images