Building Under Windows¶
howard at hobu.co
Conda contains a pre-built up-to-date 64 bit Windows binary. It is fully-featured, and if you do not need anything custom, it is likely the fastest way to get going.
Pre-built binary packages for Windows are available via Conda (64-bit version), and all of the prerequisites required for compilation of a fully featured build are also available via that packaging system. This document assumes you will be using Conda Forge as your base, and anything more advanced is beyond the scope of the document.
The AppVeyor build system uses the PDAL project’s configuration on the Conda Forge system. It contains a rich resource of known working examples. See https://github.com/PDAL/PDAL/blob/master/appveyor.yml and https://github.com/PDAL/PDAL/tree/master/scripts/appveyor for inspiration.
PDAL is known to compile on Visual Studio 2015, and 2013 might work with some source tree adjustments. PDAL makes heavy use of C++11, and a compiler with good support for those features is required.
PDAL uses the AppVeyor continuous integration platform for building and testing itself on Windows. The configuration that PDAL uses is valuable raw materials for configuring your own environment because the PDAL team must keep it up to date with both the Conda environment and the Microsoft compiler situation.
You can see the current AppVeyor configuration at
https://github.com/PDAL/PDAL/blob/master/appveyor.yml The most interesting bits
install section, the
config.cmd, and the
The AppVeyor configuration already has Miniconda installed, and the
config.cmd script installs all of PDAL’s prerequisites via the command
conda install geotiff laszip nitro curl ^ gdal pcl cmake eigen ninja libgdal ^ zstd numpy xz libxml2 laz-perf qhull ^ sqlite hdf5 tiledb conda-build ninja -y
The package list here might change over time. The canonnical location
to learn the prerequisite list for PDAL is the
file in PDAL’s source tree.
Fetching the Source¶
Get the source code for PDAL. Presumably you have GitHub for Windows or something like it. Run a “git shell” and clone the repository into the directory of your choice.
c:\dev> git clone https://github.com/PDAL/PDAL.git
Switch to the
c:\dev> git checkout 1.9-maintenance
PDAL’s active development branch is
master, and you are welcome to build it, but is not as stable as the major-versioned release branches are likely to be.
PDAL uses CMake for its build configuration. You will need to install CMake and have it available on your path to configure PDAL.
cmake command to configure the PDAL.
cmake -G "NMake Makefiles" .
A fully-featured build will require more specification of libraries, enabled features, and their locations. There are two places in the source tree for inspiration on this topic.
The AppVeyor build configuration https://github.com/PDAL/PDAL/blob/master/scripts/appveyor/config.cmd#L26
Howard Butler’s example build configuration https://github.com/PDAL/PDAL/blob/master/scripts/conda/win64.bat
Placing your command in a
.bat file will make for easy reuse.
If you chose
NMake Makefiles as your CMake generator, you can
invoke the build by calling nmake:
nmake /f Makefile
If you chose “Visual Studio 14 Win64” as your CMake generator, open
and chose your configuration to build.
After you’ve built the tree, you can run
pdal.exe by issuing it
You may need to have your Conda environment active to enable access to PDAL’s dependencies.