Installation

Installation

This page gives a quick tutorial on how to install the GDSL toolkit, based on your requirements:

• Unix (gcc or clang on Linux and MacOS)
• [Windows](##installation-on-windows] (Visual C)
• [Custom Installation](##custom-installation] for more specific needs

Installation on Unix

You need a working SML compiler/interpreter. We have tested SML/NJ and MLton. We recommend the latter which can be installed as follows:

• for Ubuntu: apt-get install mlton
• MacOS:
  • install the MLton binary pakage
  • install the dynamic library version of the GNU Multi-Precision Arithmetic library (GMP) using MacPorts (if you use fink or some other method, you need to put a link from /opt/local/lib/libgmp.10.dylib to wherever this library is installed, e.g. /sw/lib/gmp5/libgmp.10.dylib for fink)

Now download the GDSL toolkit gdsl-toolkit-x.x.x.tar.gz here.

• unpack the .tar.gz file using tar xzf gdsl-toolkit-x.x.x.tar.gz, creating a sub-directory gdsl-toolkit-x.x.x
• in this subdirectory, invoke ./configure
• do a make and (as root) make install

See the QuickStart page on how to test the resulting decoder.

Installation on Windows

The decoders can be built on Windows using a pre-build executable of the GDSL compiler.

In order to compile the demo, you need Microsoft Visual Studio and Windows SDK installed. The Visual C compiler must be on your %PATH% which is not the case by default. Go to Control Panel/System/Advanced and click Environment Variables. Here, add the path to the \bin directory of the C compiler, e.g. C:\Program Files (x86)\Microsoft Visual Studio 12.0\VC\bin. Start cmd.exe and run cl.exe to verify this.

Download the GDSL toolkit for Windows win-gdsl-toolkit-x.x.x.zip here.

• unpack the .zip file, creating a sub-directory gdsl-toolkit-x.x.x
• in this subdirectory, invoke build.bat, you can even run it from the File Explorer by double clicking the file
• open cmd.exe, change to the GDSL directory and run one of the decoders

See the QuickStart page on how to test the resulting decoder.

Custom Installation

This section details how to adjust the build process by specifying what to build and where to install. These instructions apply to the ./configure-based installation only.

The toolkit can be compiled in a directory separate from the source directory. In order to do this, call path/configure in the directory where the build should happen. Here path is the path to the configure script.

The most important options of the configure script are as follows:

• --prefix=path: installs the toolkit in path rather than in /usr/local

• --with-frontend=...: only builds the given decoders, default is the special value all. In order to see the list of available decoders, pass a dummy string to this option.

As before, the toolkit is built using

make

and (optionally) installed using

make install

See ./configure --help for more options.

Building from Repository Sources

Checking out the current sources and building them requires the following tools to be on the current search PATH:

• automake
• autoconf
• mlton
• libtool
• git

After checking out the sources, the configure script has to be generated using the following command:

autoreconf -i

• MacOS:
  • ensure that libtool is not the crippled one from Xcode; if you install a separate version but the LT_INIT macro is still not found, try setting export LIBTOOLIZE=/opt/local/bin/glibtoolize and export LIBTOOL=/opt/local/bin/glibtool

Then follow the instructions above.