Relocatable binaries¶
GROMACS (mostly) implements the concept of relocatable binaries, i.e., that
after initial installation to CMAKE_INSTALL_PREFIX
(or binary packaging
with CPack), the whole installation tree can be moved to a different folder and
GROMACS continues to work without further changes to the installation tree.
This page explains how this is implemented, and the known limitations in the
implementation. This information is mainly of interest to developers who need
to understand this or change the code, but it can also be useful for people
installing or packaging GROMACS.
A related feature that needs to be considered in all the code related to this is that the executables should work directly when executed from the build tree, before installation. In such a case, the data files should also be looked up from the source tree to make development easy.
Finding data files¶
The other, GROMACS-specific part, of making the binaries relocatable is to make them able to find data files from the installation tree. Such data files are used for multiple purposes, including showing the quotes at the end of program execution. If the quote database is not found, the quotes are simply not printed, but other files (mostly used by system preparation tools like gmx pdb2gmx and gmx grompp, and by various analysis tools for static data) will cause fatal errors if not found.
There are several considerations here:
- For relocation to work, finding the data files cannot rely on any hard-coded
absolute path, but it must find out the location of the executing code by
inspecting the system. As a fallback, environment variables or such set by
GMXRC
or similar could be used (but currently are not). - When running executables from the build tree, it is desirable that they will
automatically use the data files from the matching source tree to facilitate
easy testing. The data files are not copied into the build tree, and the
user is free to choose any relative locations for the source and build trees.
Also, the data files are not in the same relative path in the source tree and
in the installation tree (the source tree has
share/top/
, the installation treeshare/gromacs/top/
; the latter is customizable during CMake configuration). - In addition to GROMACS executables, programs that link against
libgromacs
need to be able to find the data files if they call certain functions in the library. In this case, the executable may not be in the same directory where GROMACS is. In case of static linking, no part of the code is actually loaded from the GROMACS installation prefix, which makes it impossible to find the data files without external information. - The user can always use the
GMXLIB
environment variable to provide alternative locations for the data files, but ideally this should never be necessary for using the data files from the installation.
Not all the above considerations are fully addressed by the current implementation, which works like this:
It finds the path to the current executable based on
argv[0]
. If the value contains a directory, this is interpreted as absolute or as relative to the current working directory. If there is no directory, then a file by that name is searched from the directories listed inPATH
. On Windows, the current directory is also searched beforePATH
. If a file with a matching name is found, this is used without further checking.If the executable is found and is a symbolic link, the symbolic links are traversed until a real file is found. Note that links in the directory name are not resolved, and if some of the links contain relative paths, the end result may contain
..
components and such.If an absolute path to the executable was found, the code checks whether the executable is located in the build output directory (using
stat()
or similar to account for possible symbolic links in the directory components). If it is, then the hard-coded source tree location is returned.If an absolute path to the executable was found and it was not in the build tree, then all parent directories are checked. If a parent directory contains
share/gromacs/top/gurgle.dat
, this directory is returned as the installation prefix. The file namegurgle.dat
and the location are considered unique enough to ensure that the correct directory has been found. The installation directory for read-only architecture-independent data files can be customized during CMake configuration by settingCMAKE_INSTALL_DATADIR
, and the subdirectory under this that hosts the GROMACS-specific data is set byGMX_INSTALL_DATASUBDIR
.Note that this search does not resolve symbolic links or normalize the input path beforehand: if there are
..
components and symbolic links in the path, the search may proceed to unexpected directories, but this should not be an issue as the correct installation prefix should be found before encountering such symbolic links (as long as thebin/
directory is not a symbolic link).If the data files have not been found yet, try a few hard-coded guesses (like the original installation
CMAKE_INSTALL_PREFIX
and/usr/local/
). The first guess that contains suitable files (gurgle.dat
) is returned.If still nothing is found, return
CMAKE_INSTALL_PREFIX
and let the subsequent data file opening fail.
The above logic to find the installation prefix is in
src/gromacs/commandline/cmdlineprogramcontext.cpp
. Note that code that
links to libgromacs
can provide an alternative implementation for
gmx::IProgramContext
for locating the data files, and is then fully
responsible of the above considerations.
Information about the used data directories is printed into the console output
(unless run with -quiet
), as well as to (some) error messages when locating
data files, to help diagnosing issues.
There is no mechanism to disable this probing search or affect the process during compilation time, except for the CMake variables mentioned above.
Known issues¶
GMXRC
is not relocatable: it hardcodes the absolute installation path in one assignment within the script, which no longer works after relocation. Contributions to get rid of this on all the shells theGMXRC
currently supports are welcome.- There is no version checking in the search for the data files; in case of issues with the search, it may happen that the installation prefix from some other installation of GROMACS is returned instead, and only cryptic errors about missing or invalid files may reveal this.
- If the searching for the installation prefix is not successful, hard-coded
absolute guesses are used, and one of those returned. These guesses include
the absolute path in
CMAKE_INSTALL_PREFIX
used during compilation oflibgromacs
, which will be incorrect after relocation. - The search for the installation prefix is based on the locating the
executable. This does not work for programs that link against
libgromacs
, but are not installed in the same prefix. For such cases, the hard-coded guesses will be used, so the search will not find the correct data files after relocation. The calling code can, however, programmatically provide the GROMACS installation prefix, but ideally this would work without offloading work to the calling code. - One option to (partially) solve the two above issues would be to use the
GMXDATA
environment variable set byGMXRC
as the fallback (currently this environment variable is set, but very rarely used). - Installed
pkg-config
files are not relocatable: they hardcode the absolute installation path.