> On Apr 2, 2018, at 2:00 PM, devel-requ...@lists.open-mpi.org wrote:
> 
> From: Gilles Gouaillardet <gilles.gouaillar...@gmail.com 
> <mailto:gilles.gouaillar...@gmail.com>>
> Subject: Re: [OMPI devel] Guidance on How to Submit PRs to Fix GitHub Issue 
> #5000?
> Date: April 1, 2018 at 8:49:32 PM EDT
> To: Open MPI Developers <devel@lists.open-mpi.org 
> <mailto:devel@lists.open-mpi.org>>
> 
> Bryce,
> 
> The current status on OS X is half baked, the /usr/bin/javah symlink is still 
> there, but it points to nothing, and a direct consequence is that 
> AC_PATH_PROG detects /usr/bin/javah, when I hope it does not.
> 
> ⋮


     I touch on that later on in my original issue’s discussion thread and in 
the one I opened downstream in Homebrew, yes.  I believe, however, that, if you 
haven’t noticed it already, the situation is like this:  

`AC_PATH_PROG()` finds `/usr/bin/javah` in your system’s `$PATH`.  
`/usr/bin/javah` is a symbolic link to 
`/System/Library/Frameworks/JavaVM.framework/Versions/Current/Commands/javah`.  
When Open mPI’s build process invokes 
`/System/Library/Frameworks/JavaVM.framework/Versions/Current/Commands/javah` 
through `/usr/bin/javah`, the former then serves as a run-time shim that looks 
for 
`/Library/Java/JavaVirtualMachines/jdk-$JDK_VERSION.jdk/Contents/Home/bin/javah`.
  
When 
`/System/Library/Frameworks/JavaVM.framework/Versions/Current/Commands/javah` 
can’t forward its own invocation along to 
`/Library/Java/JavaVirtualMachines/jdk-$JDK_VERSION.jdk/Contents/Home/bin/javah`
 when `$JDK_VERSION=10` because the latter tool does not exist in that case, 
the former then prints a message to standard error and then exits with error 
code 2.  

> ⋮
> 
> Since javac -h is already working in Java 8, I guess we do not care of older 
> java versions.
> 
> I am currently testing a PR that simply replace java with javac -h
> 
> ⋮


     Cool, then I’ll rebase my stab at the work that remains on top of yours 
after in lands in-tree.  

> ⋮
> 
> Regarding the PR, you are obviously free to issue one. That being said, it 
> should only be vs the master branch. Once it is merged, other PRs can be 
> issued vs the release branches by cherry-picking (and back porting if needed) 
> the commit in the master branch. This is how we work at Open MPI (e.g. no, we 
> do not gitflow).
> 
> Cheers,
> 
> Gilles

     I was already working against `master`, but thanks for letting me know.  

Regards, 
     Bryce Glover
     randomdsde...@gmail.com
_______________________________________________
devel mailing list
devel@lists.open-mpi.org
https://lists.open-mpi.org/mailman/listinfo/devel

Reply via email to