[Users] x86_64 link-time errors on MacOS Mojave

[Haas, Roland]

Hello Bernard,

thank you for the update. Glad that you could find a workaround.

These should not be needed as the scripts in ExternalLibraries/HDF5 are
supposed to figure these out based on just HDF5_DIR (and possible
pkgconfig files though MacPorts does not always deliver those). 

We had issues in the past where there were environment variables
HDF5_DIR (though never anything else really) defined in a user's
environment (for example from a environment module on a cluster) and,
due to the way make works, these take precedence over values in the
optionlists.

Can you check if you enter 

env | grep HDF5

if anything shows up, please? In particular a HDF5_DIR would likely fool
the automatic detection but as long as you *also* have the
HDF5_INC/LIB/BLAH variables set, then HDF5_DIR is (almost, b/c it
is used to find things like h5ls) not needed.

Yours,
Roland

> Hi Roland, everyone:
> 
> I just came back to this problem in the last day or two, and with some
> experimentation, I fixed it by adding the following lines to my
> config-info ---that is, to
> simfactory/mdb/optionlists/osx-macports.cfg:
> 
> HDF5_INC_DIRS=/opt/local/include
> HDF5_LIBS=hdf5 hdf5_cpp hdf5_hl
> HDF5_LIB_DIRS=/opt/local/lib
> 
> That is, it wasn't enough to just say "HDF5_DIR=/opt/local"
> 
> Hope this helps if someone else runs into problems (and thanks for the
> help getting here),
> 
> Bernard
> 
> On Fri, 21 Jun 2019 at 14:20, Haas, Roland <rhaas at illinois.edu> wrote:
> >
> > Hello Bernard,
> >
> > unfortunately (or fortunately) it linked fine for me. No errors were
> > reported.
> >
> > Looking at the installed ports I also cannot see anything strange.
> >
> > I am not sure what else I could now do to help debug this. Possibly
> > some other package that you have installed is interfering or there is
> > some environment variable set?
> >
> > Yours,
> > Roland
> >  
> > > Hi Roland.
> > >
> > > Here's the result of uname -a:
> > >
> > > Darwin gs66-dodder 18.6.0 Darwin Kernel Version 18.6.0: Thu Apr 25
> > > 23:16:27 PDT 2019; root:xnu-4903.261.4~2/RELEASE_X86_64 x86_64
> > >
> > > Attached is the full list of installed ports (a bit longer than yours)
> > >
> > > Bernard
> > >
> > > On Thu, 20 Jun 2019 at 12:44, Haas, Roland <rhaas at illinois.edu> wrote:  
> > > >
> > > > Hello Bernard,
> > > >  
> > > > > I did a clean, from-scratch compilation after copying that 18.2.0
> > > > > Darwin architecture file to 18.6.0.
> > > > >
> > > > > Sadly, no change. Still fails at link time with all the "Undefined
> > > > > _H5xxxx" errors.  
> > > > Ok. Since it starts with at "_" I almost suspected it may be some
> > > > Fortran binding but then it is being called from C++ code.
> > > >
> > > > My own test is still compiling (slow VM).
> > > >
> > > > Could you include the output of:
> > > >
> > > > * sudo port installed
> > > > * uname -a
> > > >
> > > > I have included my own output to give you an idea of what I am looking
> > > > for (note that I cannot tell yet if compiling in my VM will encounter
> > > > the same error you see).
> > > >
> > > > Yours,
> > > > Roland
> > > >
> > > > --
> > > > My email is as private as my paper mail. I therefore support encrypting
> > > > and signing email messages. Get my PGP key from http://pgp.mit.edu .  
> > >
> > >
> > >  
> >
> >
> >
> > --
> > My email is as private as my paper mail. I therefore support encrypting
> > and signing email messages. Get my PGP key from http://pgp.mit.edu .  
> 
> 
> 



-- 
My email is as private as my paper mail. I therefore support encrypting
and signing email messages. Get my PGP key from http://keys.gnupg.net.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 833 bytes
Desc: OpenPGP digital signature
Url : http://lists.einsteintoolkit.org/pipermail/users/attachments/20190702/2f153ff8/attachment.bin