Release of PnetCDF 1.9.0

[Wei-keng Liao]

Could you try the patch below (only one line changed)?
I found it a better solution than adding LDFLAGS.


--- src/libs/Makefile.am	(revision 3766)
+++ src/libs/Makefile.am	(working copy)
@@ -44,7 +44,7 @@
 libpnetcdf_la_LIBADD += ../binding/f90/libf90.la
 # According to Automake Manual, Section 8.14.3 Mixing Fortran 77 With C and
 # C++, adding $(FLIBS) is necessary
-# libpnetcdf_la_LIBADD += $(FLIBS) $(FCLIBS)
+libpnetcdf_la_LIBADD += $(FLIBS) $(FCLIBS)
 # nodist_EXTRA_libpnetcdf_la_SOURCES += dummyf90.f90
 endif
 


Wei-keng

> On May 19, 2018, at 12:28 PM, Wei-keng Liao <wkliao at eecs.northwestern.edu> wrote:
> 
> Hi, Bruno
> 
> I am wondering if you still have the same problem.
> 
> If yes, could you try adding the following environment variable to your configure command line.
> LDFLAGS="-Wl,--allow-shlib-undefined”
> 
> The error messages you encountered also appear on other platforms, Ubuntu and Redhat.
> See doc/README.Ubuntu from the release for more info.
> 
> Please let me know whether this works for ArchLinux.
> 
> Wei-keng
> 
>> On Dec 20, 2017, at 5:13 PM, Wei-keng Liao <wkliao at eecs.northwestern.edu> wrote:
>> 
>> 
>> If GNU 7.2.1 is a suspect, would you like to try an earlier
>> version of GNU compilers to verify?
>> 
>> Wei-keng
>> 
>> On Dec 20, 2017, at 5:01 PM, Bruno Pagani wrote:
>> 
>>> Le 20/12/2017 à 23:50, Wei-keng Liao a écrit :
>>>> I just built an OpenMPI 3.0.0 with GNU 6.2.0 compilers on a Redhat but
>>>> could not reproduce the errors you are seeing.
>>> 
>>> The issue arise with both OpenMPI 2.1.1 and 3.0.0. I would rather expect
>>> GNU 7.2.1 to be the culprit here.
>>> 
>>>> To check whether the problem is OpenMPI related, could you try
>>>> using MPICH and see if the errors go away?
>>> 
>>> Will take me some time as MPICH is not currently available on ArchLinux
>>> (this is on my TL, but not reachable in a foreseeable future), so I
>>> won’t be able to test it within our building system. But if I can build
>>> it locally and try with that maybe that could do it (I would need to
>>> reproduce the error outside our building system first, but that should
>>> be straight away).
>>> 
>>> I’ll keep you updated, but expect some weeks before I do.
>>> 
>>> Regards,
>>> Bruno
>> 
>