2.1.1 Compiling the code

vertical layer

2023-03-17T03:05:44Z

Hi, i use the delft3d fm 2022, i want to set different vercial layers. i do as the manual said. i add "StretchType=1"and"StretchCoef=4.0, 5.9, 8.7, 12.7, 18.7, 18.7, 12.7, 8.7, 5.9, 4.0" in the mdu file,but it doesnt work. how to set different layers?

next Linux distribution and version

2023-01-27T13:51:31Z

the get-started pages (Delft3D and Delft3D-FM) say that CentOs7 or RedHat7 is required as linux version.
Red Hat Linux 7 is end-of-life for years already (or was RHEL7 meant there?), CentOs7 will be maintained till 2024 but will not have a successor.

One of our researchers has to start with Delft3D in Linux and would like to continue with that installation for the coming years. What would be the best distribution and version to start with now and where to upgrade to in future?

Thanks,
Roel, IHE Delft

Building in Windows 10 with Visual Studio 2022 and Intel oneAPI 2022

2023-01-25T17:41:22Z

Hello!

I was able to build Delft3D FM (141208) with no problem in Windows 10 with Visual Studio 2022 and Intel oneAPI 2022

{root}/build.bat:

Lines 195 to 200:

1	if NOT "%VS2022INSTALLDIR%" == "" (
2	set vs=2022
3	echo Found: VisualStudio 17 2022
4	if NOT "%IFORT_COMPILER22%" == "" (
5	set ifort=22
6	echo Found: Intel Fortran 2022

Lines 312 and 313:

1	if "!vs!" == "2022" (
2	set generator="Visual Studio 17 2022"

Lines 357 to 359:

1	if !generator! == "Visual Studio 17 2022" (
2	echo "Calling vcvarsall.bat for VisualStudio 2022 ..."
3	call "%VS2022INSTALLDIR%\VC\Auxiliary\Build\vcvarsall.bat" amd64

For other projects and Delft3D FM I have the following VS 2022 workloads installed:

1	Microsoft.VisualStudio.Workload.CoreEditor
2	Microsoft.VisualStudio.Workload.NativeDesktop
3	Microsoft.VisualStudio.Component.VC.Tools.x86.x64
4	Microsoft.VisualStudio.Component.VC.DiagnosticTools
5	Microsoft.VisualStudio.Component.Windows11SDK.22621
6	Microsoft.VisualStudio.Component.VC.CMake.Project
7	Microsoft.VisualStudio.Component.VC.ATL
8	Microsoft.VisualStudio.Component.VC.ATLMFC
9	Microsoft.VisualStudio.Component.VC.CLI.Support
10	Microsoft.Component.VC.Runtime.UCRTSDK

Downloaded and installed Intel:

Fortran Compiler
MPI
MKL

ENVs:

I_MPI_ONEAPI_ROOT = C:\Program Files (x86)\Intel\oneAPI\mpi\2021.7.1
IFORT_COMPILER22 = C:\Program Files (x86)\Intel\oneAPI\compiler\2022.2.1\windows\
ONEAPI_ROOT = C:\Program Files (x86)\Intel\oneAPI\
VS2022INSTALLDIR = C:\Program Files\Microsoft Visual Studio\2022\Community

{root}/build.bat all was enough

When I try oneAPI 2023, it seems that it tries to copy files from a folder {root}/src/third_party_open/intelredist/ and doesn't copy important libs like impi.dll

Running suggested example results in: symbol lookup error undefined symbol

2023-01-11T07:47:01Z

On linux after compile. When trying to run the suggested example <myCode>/examples/12_dflowfm/test_data/e100_f02_c02-FriesianInlet_schematic_FM the error message symbol lookup error: <myCode>/build_all/lnx64/bin/../bin/dimr: undefined symbol: __libm_feature_flag_x pops up. Does anyone know what is going on here?
From file size alone it is clear that libimf.so in lnx64/lib is not copied from one of the files in the tree under /opt/intel/oneapi. Could that be the cause?

Found the probable cause: I am compiling netcdf with the intel compilers. Reason: the gnu compiler suite version used in versions 8 or 9 of Redhat clones to build epel packages is not one of those for which the intel compiler suite recognizes the netcdf module file. Going back to the tests for netcdf-c I spotted a failure with the same symbol problem. While researching this I found a website stating that hdf5 needs to be compiled with the same compiler as netcdf. Doing so resolved this particular problem.

On linux Intel 2023 adds errors

2023-01-09T12:34:41Z

The new 2023 Intel OneAPI produces errors. Error numbers are #7938 and #7983. These errors seem to spring from a conflict between a variable c_net_file_in being declared of fixed length
```
character(kind=c_char), intent(in) :: c_netfile_in(MAXSTRLEN)
```
and a variable netfile_in being declared with length derived from c_netfile_in

```
character(len=strlen(c_netfile_in)) :: netfile_in
```
while making c_netfile_in variable length removes the error messages, I have not checked whether that breaks things. Does anyone know?

RE: The argument to C_LOC must be a variable with POINTER or TARGET attr

2023-01-09T12:10:13Z

In principle adding target to the definitions on lines 109 and 116 of src/engines_gpl/dflowfm/packages/dflowfm_kernel/src/dflowfm_data/m_flowexternalforcings.f90 makes the error go away. Moreover, given that adding target only means the varieable might be a pointer target, this shuodl not cause other problems.

RE: Notes on install delft3fm (tag version 141346) on windows laptop

2022-12-26T18:37:03Z

Hi youngling,

thanks for asking, I just updated my notes and I hope this new version is more clear and useful to you.

I have not seen the error you met before so sorry I don't know what might be the cause. I would suggest to start with a clean system and install everything step by step, make sure you install all components needed when installing visual studio, I think this is where things normally go wrong cause some components might be missed.

since you are using oneapi 2023, I would suggest to pay more attention to the part about oneapi 2022 in my notes and take necessary actions when you install with oneapi 2023.

hope it goes well with your installation.

Cheers,
Zhiling

RE: Notes on install delft3fm (tag version 141346) on windows laptop

2022-12-23T09:08:30Z

Thanks for your explanation.
in my case, i used intel oneapi 2023(other is same).
I followed your method exactly but got an error(please check under cc3.png file)
(also alot of warnings)

i guess that intel oneapi 2023 is something different from 2021.
do you have any idea about solveing this ploblem.

The argument to C_LOC must be a variable with POINTER or TARGET attr

2022-12-19T07:47:06Z

I am working on compilation of Delft3D-FM on OracelLInux 8.7 with the intel oneapi 2022.3 base and HPC tools.
The Intel compiler from oneapi 2022.3 gives the following error messages for the delft3dfm tag 142074 source code
```
src/engines_gpl/dflowfm/packages/dflowfm_lib/src/unstruc_dll_api.F90(104): error #9022: The argument to C_LOC must be a variable with the POINTER or TARGET attribute.   [NBNDZ]
      x = c_loc(nbndz)
----------------^
src/engines_gpl/dflowfm/packages/dflowfm_lib/src/unstruc_dll_api.F90(108): error #9022: The argument to C_LOC must be a variable with the POINTER or TARGET attribute.   [KBNDZ]
      x = c_loc(kbndz)
----------------^

```
A search on internet suggests this is new behavior in the intel 2022.3 version of the compiler. Does anyone know a work-around for this?

Notes on install delft3fm (tag version 141346) on windows laptop

2022-12-26T18:27:09Z

Hi everyone,

i figured out how to build from source code to install delft3fm on my windows laptop and just want to share my notes of the key steps and how to fix bugs. See the attached .html file. in order to open the .html normally, put the materials (.png and .css) into a folder with the same name of the .html file.

Hope it helps.

Zhiling

Anybody successfully build on Windows 11?

2022-11-06T22:15:47Z

Hi everyone,

I plan to order a PC to test cases for both delft3d 4 and FM, but it seems hard find a windows 10 PC on market nowadays. I consulted the software@deltares they said it's not officially tested on windows 11, so just want to ask has anyone successfully built the code (4 or FM suite) on windows 11? anything particular to notice?

Thank you very much.
Regards,
Zhiling

OpenMP not available (LINUX)

2022-10-04T03:02:19Z

Hello,

I am trying to run simulations in LINUX (RedHat) by utilising the Singularity container. After running the simulations it is reported by the .dia (diagnosis) file that OpenMP is not available. I would like to know why is this issue happening and how could I set the number of threads as input (number of cores utilised). Thank you in advance.

[dflowfm_69918] Silent errors during build

2022-09-02T15:13:16Z

HI,

I have been unsuccessfully trying to builf the dflowFM program for a while now. I must confess that I am not familiar with CMake. The furthest I came to building the program is this instance where the script (src/build_gnu.sh) says that the build was finished but in fact, the binaries were not build in src/bin. On inspection of the make logfile, I noticed a few silent errors (complete logfiles can be found as an attachment):

```
( \
cd . ; \
./install.sh /home/dflowfm/dflowfm_69918/src \
)
oss-install...
Prefix : /home/dflowfm/dflowfm_69918/src
Target directory : /home/dflowfm/dflowfm_69918/src/bin
Current directory : /home/dflowfm/dflowfm_69918/src
Source root dir : /home/dflowfm/dflowfm_69918/src/scripts_lgpl/linux/../..
Gathering dependent libraries . . .
Gathering libraries for lib/* ...
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
cp: missing destination file operand after '/home/dflowfm/dflowfm_69918/src/lib'
Try 'cp --help' for more information.
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
Gathering libraries for bin/* ...
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
cp: missing destination file operand after '/home/dflowfm/dflowfm_69918/src/lib'
Try 'cp --help' for more information.
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
not a dynamic executable
```

System configuration:
- Ubuntu 22.04 on WSL2 (x64)
- Intel OneAPI 2022.06

libtool_install.sh is not generated to continue further with the instructions provided.

Any help is highly appreciated. I am stuck trying to build (even tried tag:68819 as suggested in the readme) for a couple of weeks now! And yes, I tried to follow the steps mentioned in src/README.md as well (by exporting flags as necessary petsc and netcdf_fortran). I have even tried compiling it on CentOS7 (tag:68819), however, the older version of gfortran packaged on it throws some errors.

Is there perhaps a dockerised version the program or better documentation someone can point me to? That would be great.

Thanks in advance!
Cheers,
Veethahavya

RE: Error retrieving FM Dll version: Could not find / load dflowfm.dll

2022-07-22T20:36:34Z

The problem is solved. The current build.bat script only has the build settings for Intel oneAPI 2021. However, Intel has updated the complete oneAPI to a 2022.1 version. Setting the prepare.py as described here solved the problem: Compiling Delft3D FM with Visual Studio 2019 and Intel oneAPI (2021) - YouTube.

Error retrieving FM Dll version: Could not find / load dflowfm.dll

2022-07-04T11:07:11Z

I've compiled the source code as written in the get started page for FM, without errors and successfully run the example files.
The "build_all" folder with the x64 subdirectory (am using windows 10 enterprise, intel procc.) is created as it should with all the kernel libraries compiled.

I´ve then copied all the compiled kernels from C:\delft3dfm_rep\build_all\x64\Release into the GUI folder: C:\Program Files\Deltares\Delft3D FM Suite 2022.03 OpenHMWQ\plugins\DeltaShell.Dimr\kernels\x64.

Still, I am getting the following error when running the FM Suite OpenHMWQ(2022.03) version:

Does anyone have an idea what the problem could be?

I´ve made sure the dflowfm.dll module is in fact in the directory the error is referring to:

Still I am getting the error when running FM Suite and cannot run a model. Everything else seems to be working fine.

RE: The build scripts stuck at the second last line of 'install_and_bundle

2022-05-31T14:06:04Z

See the duplicate thread

The build scripts stuck at the second last line of 'install_and_bundle.'

2022-05-27T05:07:53Z

system: deepin 20.5 (A debian based linux);
compiler: intel oneapi 2021.02.
I add some codes in the build.sh as below:

############add to build.sh after "On Deltares systems only"#######
# Intel compiler:
myconfig=$config
. /opt/intel/oneapi/setvars.sh
. /opt/intel/oneapi/tbb/latest/env/vars.sh
export config=$myconfig

# Intel MPI:
. /opt/intel/oneapi/mpi/latest/env/vars.sh -ofi_internal=1

# netcdf、petsc:
export NETCDF=~/oss_dependence
export PKG_CONFIG_PATH=${NETCDF}/pkg_config:$PKG_CONFIG_PATH

# Metis:
export METIS_DIR=${NETCDF}/metis

export FC=mpiifort
export CXX=mpiicpc
export CC=mpiicc
echo "FC=$FC"
echo "CXX=$CXX"
echo "CC=$CC"

#########add code end###########
It seems good, but stucked at the second last line of 'install_and_bundle.cmake':
"execute_process(COMMAND find "${CMAKE_INSTALL_PREFIX}/lib" -type l -exec echo "remove destination of symlink:" {} \; -exec bash -c "cp --remove-destination $(readlink {}) {};" {} \; WORKING_DIRECTORY "${CMAKE_INSTALL_PREFIX}/lib" )"

And , It cannot find the folder "${CMAKE_INSTALL_PREFIX}/share".
Could you help me? Thanks very much!