[Dune] using metis & parmetis creates a build problem

Benedikt Oswald benedikt.oswald at lspr.ch
Fri May 2 12:34:42 CEST 2014


Hello Markus, thanks!

parmetis version 4.0.3/openmpi/1.6.5/gcc/4.8.1

Greetings, Benedikt






--------------------------------------------------------------------------------------------------------------------------------------------------------------
Dr. sc. techn. Benedikt Oswald - first engineer - LSPR AG - phone - +41 43 366 90 74
Technoparkstrasse 1, CH-8005 Zürich, benedikt.oswald at lspr.ch - labor vincit omnia improbus
--------------------------------------------------------------------------------------------------------------------------------------------------------------

On 02 May 2014, at 12:33, Dr. Markus Blatt <markus at dr-blatt.de> wrote:

> On 1. Mai 2014 09:16:30 MESZ, Benedikt Oswald <benedikt.oswald at lspr.ch> wrote:
>> Hello Markus,  thanks for the fix. In fact, the metis.h not found error
>> goes away,
>> but there is another problem with istl:
>> 
>> ===========
>> In file included from
>> /Users/oswald/gitwork/hades-pure-dune/dune-istl/dune/istl/matrixredistribute.hh:5:0,
>> from
>> /Users/oswald/gitwork/hades-pure-dune/dune-istl/dune/istl/paamg/transfer.hh:7,
>> from
>> /Users/oswald/gitwork/hades-pure-dune/dune-istl/dune/istl/paamg/amg.hh:10,
>>                from pamgtest1.cc:23:
>> /Users/oswald/gitwork/hades-pure-dune/dune-istl/dune/istl/repartition.hh:
>> At global scope:
>> /Users/oswald/gitwork/hades-pure-dune/dune-istl/dune/istl/repartition.hh:791:67:
>> error: ‘idxtype’ has not been declared
>> inline bool isValidGraph(std::size_t noVtx, std::size_t gnoVtx, idxtype
>> noEdges, idxtype* xadj,
>>                                                                  ^
>> /Users/oswald/gitwork/hades-pure-dune/dune-istl/dune/istl/repartition.hh:791:84:
>> error: ‘idxtype’ has not been declared
>> inline bool isValidGraph(std::size_t noVtx, std::size_t gnoVtx, idxtype
>> noEdges, idxtype* xadj,
>>                                                                     ^
>> /Users/oswald/gitwork/hades-pure-dune/dune-istl/dune/istl/repartition.hh:792:28:
>> error: ‘idxtype’ has not been declared
>>                           idxtype* adjncy, bool checkSymmetry)
>>                           ^
>> /Users/oswald/gitwork/hades-pure-dune/dune-istl/dune/istl/repartition.hh:
>> In function ‘bool Dune::isValidGraph(std::size_t, std::size_t, int,
>> int*, int*, bool)’:
>> /Users/oswald/gitwork/hades-pure-dune/dune-istl/dune/istl/repartition.hh:796:9:
>> error: ‘idxtype’ was not declared in this scope
>>    for(idxtype vtx=0; vtx<(idxtype)noVtx; ++vtx) {
>>        ^
>> /Users/oswald/gitwork/hades-pure-dune/dune-istl/dune/istl/repartition.hh:796:17:
>> error: expected ‘;’ before ‘vtx’
>>    for(idxtype vtx=0; vtx<(idxtype)noVtx; ++vtx) {
>>                ^
>> /Users/oswald/gitwork/hades-pure-dune/dune-istl/dune/istl/repartition.hh:796:24:
>> error: ‘vtx’ was not declared in this scope
>>    for(idxtype vtx=0; vtx<(idxtype)noVtx; ++vtx) {
>>                       ^
>> /Users/oswald/gitwork/hades-pure-dune/dune-istl/dune/istl/repartition.hh:796:37:
>> error: expected ‘;’ before ‘noVtx’
>>    for(idxtype vtx=0; vtx<(idxtype)noVtx; ++vtx) {
>>                                    ^
>> /Users/oswald/gitwork/hades-pure-dune/dune-istl/dune/istl/repartition.hh:796:42:
>> error: expected ‘)’ before ‘;’ token
>>    for(idxtype vtx=0; vtx<(idxtype)noVtx; ++vtx) {
>>                                         ^
>> /Users/oswald/gitwork/hades-pure-dune/dune-istl/dune/istl/repartition.hh:796:46:
>> error: ‘vtx’ was not declared in this scope
>>    for(idxtype vtx=0; vtx<(idxtype)noVtx; ++vtx) {
>> =============
>> 
>> In the opts file I use both metis and parmetis. I remember, a few
>> months ago, there was something sounding similar ?
>> 
>> 
>> platform is: mac os x, gcc 4.8.2, openmpi 1.8, latest Dune from trunk
>> 
>> 
>> 
>> Greetings, Benedikt
>> 
>> 
>> 
>> 
>> 
>> 
>> 
>> On 29 Apr 2014, at 10:50, Markus Blatt <markus at dr-blatt.de> wrote:
>> 
>>> Hi,
>>> 
>>> thanks for the report. There were still some hickups in the fix and
>>> the compile flags which are hopefully fixed by now. It might be worth
>>> trying again.
>>> 
>>> Markus
>>> -- 
>>> Do you need more support with DUNE or HPC in general? 
>>> 
>>> Dr. Markus Blatt - HPC-Simulation-Software & Services
>> http://www.dr-blatt.de
>>> Hans-Bunte-Str. 8-10, 69123 Heidelberg, Germany
>>> Tel.: +49 (0) 160 97590858
> 
> Hi,
> Thanks for the feedback. Might have broken this one and will try to fix this. What version of parmetis are you using?
> 
> Markus
> -- 
> Dr. Blatt HPC-Simulation-Software & Services
> Hans-Bunte-Str. 8-10, 69123 Heidelberg, Germany

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.dune-project.org/pipermail/dune/attachments/20140502/71b790d2/attachment.htm>


More information about the Dune mailing list