[wersoe]

Hi,

why did the developers remove QT in ThirdParty-3.0?
Now, we cant compile up-to-date QT for the compilation of paraview...

Any reasons for that, otherwise: please add it again...

Sören

[zordiack]

I think you should be able to still use the makeQt build script available in ThirdParty-2.3.x and get the Qt sources from https://download.qt.io/archive/qt/4.8/4.8.6/. Also beware of this bug: https://bugreports.qt.io/browse/QTBUG-5385. This should give you ThirdParty Qt as before.

EDIT: You also need to pull etc/tools/QtFunctions and maybe etc/relocateQt

I'm just trying to compile it myself, I'll report back..

[wyldckat]

Quick note: I've made the request to bring it back on this bug report: http://www.openfoam.org/mantisbt/view.php?id=1905

[zordiack]

I also think it should be included. Especially cluster environments can have rather limited/old software support. That's why I'm currently trying to compile it myself.

As a side note, it would be nice if there was a script to easily compile also ThirdParty gmp/mpfr libs. This can be achieved using the makeGcc build script, if you comment out the check for gcc version and only provide gmp and mpfr sources it will build and install those first and then fail. It's not as pretty as it could be though

[wyldckat]

Originally Posted by zordiack

As a side note, it would be nice if there was a script to easily compile also ThirdParty gmp/mpfr libs. This can be achieved using the makeGcc build script, if you comment out the check for gcc version and only provide gmp and mpfr sources it will build and install those first and then fail. It's not as pretty as it could be though

Quick answer: That might be harder to convince Henry, but feel free to report this on the bug tracker! Splitting makeGcc into to 2 scripts isn't very hard... or even to simply add the optional arguments "-skipGCC" and "-skipMPC"

[wyldckat]

FYI, Henry has committed the changes to ThirdParty 3.0.x: https://github.com/OpenFOAM/ThirdPar...fe2209741caff2

In addition, please check his comment about issues with Qt and RHEL 6: http://www.openfoam.org/mantisbt/view.php?id=1905#c5583

[zordiack]

I can confirm that I was finally able to build ParaView-4.4.0 against ThirdParty Qt-4.8.6 sources using the makeQt script. However, there was a lot of trouble with compiling the PV4 readers (which is probably totally unrelated to Qt compilation, but I'll report it anyway). ck,

EDIT2: Did a remove-all-rebuild and now it's working

The version detection in etc/tools/ParaView4Functions fails, resulting in a complaint about version mismatch. This function returns empty on my cluster environment (Scientific Linux 6.x?):

getVersion()
{
    (
        set -- $(
            sed -ne 's/^SET(PARAVIEW_VERSION_[A-Z]*\([ 0-9]*\)).*$/\1/p' \
            $ParaView_SOURCE_DIR/CMakeLists.txt
        )

        IFS='.'
        echo "$*"
    )
}

leading to failure at the following version check.

The more serious error is that for some reason ParaView-4.4.0 source version check also fails and detects source version 3.0.0 (?). Because of this, the ParaView include files are installed in the wrong place under ThirdParty-x/platforms/linux64x/ParaView-4.4.0/include. There is now a directory paraview-3.0 because of the failed source version detection.

PV4Readers link agains $ParaView_INCLUDE_DIR, which is correctly set to ../paraview-4.4. This leads to build failure of the PV4Readers.

As a dirty fix, I commented out the git version check in file ParaView-4.4.0/CMake/ParaViewDetermineVersion.cmake. It's not a correct thing to do, but now the source version gets a correct value. Like this:

  if (NOT PARAVIEW_GIT_DESCRIBE)
  #  if (EXISTS ${git_command})
  #    execute_process(
  #      COMMAND ${git_command} describe
  #      WORKING_DIRECTORY ${source_dir}
  #      RESULT_VARIABLE result
  #      OUTPUT_VARIABLE output
  #      ERROR_QUIET
  #      OUTPUT_STRIP_TRAILING_WHITESPACE
  #      ERROR_STRIP_TRAILING_WHITESPACE)
  #  endif()
  else()
    set(result 0)
    set(output ${PARAVIEW_GIT_DESCRIBE})
  endif()

This needs a bug report, I'll create one later when I have the time to determine relevant software versions. I'm using ThirdParty-dev, but this also applies to ThirdParty-3.x I suppose.

PS. About the comment Henry made about RHEL and Qt versions, I totally agree but that needs to be told to the system administrators I would very much like to update A LOT of software on the cluster environments I'm using, but it's not my decision to make. I think many other cluster users also face the same situation (the missing sudo access problem). This is why ThirdParty compilation is a nice option to have if you want to use the latest ParaView for remote postprosessing (I know I do).

[wyldckat]

Greetings Pekka,

I'm a bit confused about the errors you're getting. Perhaps you mistyped and used makeParaView instead of makeParaView4?

What about the build log for ParaView? Did you get something like the following near the start of the output?

---------------------
Version information
    qt        4.8.1
    version   4.4.0
    major     4.4
    build     Release
MISMATCH!
    specified 4.4.0
    found     
---------------------

Because this is normal. I'm currently trying to do a battery of builds for adding the detailed installation instructions for openfoamwiki.net, so I'll be revising and testing the instructions for CentOS 6.7. I'll have them up as soon as I can.


In the meantime, there is something that comes to mind: perhaps you're getting a conflict of Qt versions, like Henry warned about? Try using ldd to give you some details on this. For example:

ldd $ParaView_DIR/lib/paraview-4.4/paraview

Check if the Qt libraries are being loaded from the system or from the custom build.

Best regards,
Bruno

[zordiack]

Ok I didn't know the first error is "normal" The source detection error however is something I've never encountered before. I don't think it has anything to do with Qt, since it is from ParaView sources and only uses git as far as I can tell. It outputs the line

Determined Source Version : 3.0.0

although the earlier version display shows 4.4.0. I'm also pretty sure I used the correct build script. I've never had this problem with Ubuntu based distros, this happened in cluster environment (Scientific Linux).

After all I got it all working, PV4readers and ParaView, after I commented the check out. I also checked ldd and it's pulling the correct Qt libs (from ThirdParty), the system version here is too old for compiling.

[wersoe]

Hi all,

sorry for being back to this quite late... I was not around at weekend.

I was able to compile Paraview 4.4 just with using my old qt-4.8.3 from OpenFOAM-2.4.x
There was no trouble at all.

I ll give you the install tomorow when I am back in office...

By the way, for me, at Rocks 6 cluster and other "older" RHEL 6 systems:
I allways use the ThirdParty compiler, means gcc-4.9.2 in the case of OF24 and OF30 with all the libraries as well as cmake and qt from ThirdParty. There is an issue withn one of the versions, Henry told me once... dont remember it. So, I always use gcc-4.9.2 with libraries suggested by OF, cmake suggested by OF and qt-4.8.3.

Best, Sören

[akidess]

Slightly off-topic: What's the reason for compiling Paraview? I've found the static binaries from Kitware to work great.

[zordiack]

MPI and Python support for me, I use ParaView a lot in client-server mode, because cases are often so large it's not practical or possible to postprocess them on desktop.

[wersoe]

OK, as promised some more info:

First, thanks Henry for putting qt back to ThirdParty.
Some details: I never saw issues with qt and different versions. However, I need to add, that I don't use it at all besides compiling paraview, since OpenFOAM is almost the very only application running on that machine...
I do need RHEL6, since Rocks Cluster based on it. There is no new version based on RHEL7, and even if it comes soon, probably no atll users will upgrade a cluster easily...

What do I do to install it at RHEL, done for 2.4 and for 3.0 (for 3.0 just use the old qt installation for paraview, I know it's dirty, but it works...)

Install packages as describes at OpenFOAM homepage.

cd /share/apps/
mkdir OpenFOAM
cd OpenFOAM/
git clone git://github.com/OpenFOAM/OpenFOAM-2.4.x.git
cd OpenFOAM-2.4.x; git pull; cd ..
wget http://downloads.sourceforge.net/foa...arty-2.4.0.tgz
tar xzf ThirdParty-2.4.0.tgz; mv ThirdParty-2.4.0 ThirdParty-2.4.x
cp Pakete-ThirdParty/* ThirdParty-2.4.x/


gedit OpenFOAM-2.4.x/etc/bashrc
--- change installation location to /share/apps/...
--- change compiler to "ThirdParty"
--- change compiler version to Gcc49
gedit OpenFOAM-2.4.x/etc/config/settings.sh
--- use Gcc 4.9.2
--- check versions of Gcc and corresponding packages, mpfr, mpc, gmp
gedit ThirdParty-2.4.x/makeGcc
--- check versions of Gcc and corresponding packages, mpfr, mpc, gmp
gedit ThirdParty-2.4.x/makeCmake
--- check version of cmake
gedit ThirdParty-2.4.x/makeQt
--- change version to 4.8.3
gedit ThirdParty-2.4.x/makeParaView4
--- set PATH to qt
--- # Set the path to the Qt-4.5 (or later) qmake if the system Qt is older
--- QMAKE_PATH="/share/apps/OpenFOAM/ThirdParty-2.4.x/platforms/linux64Gcc49/qt-4.8.3/bin/"
--- # Set the path to cmake
--- CMAKE_PATH="/share/apps/OpenFOAM/ThirdParty-2.4.x/platforms/linux64Gcc49/cmake-2.8.12.1/bin/"
gedit OpenFOAM-2.4.x/etc/config/CGAL.sh
--- change boost to boost_1_55_0
gedit ThirdParty-2.4.x/makeCGAL
--- cgalPACKAGE=CGAL-4.6
--- boostPACKAGE=boost_1_55_0
--- gmpPACKAGE=gmp-5.1.2
--- mpfrPACKAGE=mpfr-3.1.2
gedit OpenFOAM-2.4.x/etc/config/metis.sh
--- check versions

### git make files not changable
vi OpenFOAM-2.4.x/.gitignore
# OWN CONFIG FILES
/etc/bashrc
/etc/config/settings.sh
/etc/config/CGAL.sh
/etc/config/metis.sh
###

### START ###
cd ThirdParty-2.4.x/

wget ftp://ftp.mpi-sb.mpg.de/pub/gnu/mirr...c-4.9.2.tar.gz && tar xfz gcc-4.9.2.tar.gz && wget http://ftp.gnu.org/gnu/gmp/gmp-5.1.2.tar.bz2 && tar xf gmp-5.1.2.tar.bz2 && wget http://www.multiprecision.org/mpc/do...c-1.0.1.tar.gz && tar xfz mpc-1.0.1.tar.gz && wget http://www.mpfr.org/mpfr-3.1.2/mpfr-3.1.2.tar.gz && tar xfz mpfr-3.1.2.tar.gz && wget http://www.mirrorservice.org/sites/d...c-4.8.3.tar.gz && tar xfz qt-everywhere-opensource-src-4.8.3.tar.gz && wget http://downloads.sourceforge.net/pro..._1_55_0.tar.gz && tar xfz boost_1_55_0.tar.gz && wget http://glaros.dtc.umn.edu/gkhome/fet...s-5.1.0.tar.gz && tar xfz metis-5.1.0.tar.gz

. ../OpenFOAM-2.4.x/etc/bashrc && ./makeGcc 2>&1 | tee log.makeGcc

. ../OpenFOAM-2.4.x/etc/bashrc && ./makeCmake | tee log.makeCmake

. ../OpenFOAM-2.4.x/etc/bashrc && ./makeQt 2>&1 | tee log.makeQt

. ../OpenFOAM-2.4.x/etc/bashrc && ./Allwmake 2>&1 | tee log.Allwmake

. ../OpenFOAM-2.4.x/etc/bashrc && ./makeParaView4 2>&1 | tee log.makeParaView4


export WM_SCHEDULER=wmakeScheduler
export WM_HOSTS="localhost:4"
export WM_NCOMPPROCS=$($WM_SCHEDULER -count)
###

cd ../OpenFOAM-2.4.x/ && . etc/bashrc && ./Allwmake 2>&1 | tee log.Allwmake

### add to ~/.bashrc
alias OF24='. /share/apps/OpenFOAM/OpenFOAM-2.4.x/etc/bashrc && . /share/apps/OpenFOAM/OpenFOAM-2.4.x/bin/tools/RunFunctions'