As discussed in the introduction, the main way of distributing and obtaining CREDO is as a core part of the stgUnderworld geophysics framework.
So setting up to use CREDO is very similar to setting up stgUnderworld, except for some extra Python configuration at the conclusion.
CREDO has been designed to operate with just Python as a core dependency - other libraries for more advanced functionality are optional, although are recommended to use the full functionality.
CREDO is written to work with Python 2.5 or 2.6. One of these versions should come pre-installed on most Linux systems (including supercomputees), Mac OS machines, and clusters. For Windows, you may need to download and install Python yourself - check out the Python website for instructions and documentation.
For users with a Mac machine or intending to run analysis on a cluster, you may wish to consider one of the “scientific Python” distributions, which will install Python, plus a large set of useful scientific libraries including those mentioned below, as a group in one large install package. The idea is this takes away some of the pain of individually installing these libraries, and making sure you have compatible versions installed. Some of the leading “install sets” of this type are:
we don’t expressly recommend either of these packages at this time, but provide them as alternative options to the regular install process. As mentioned above, they may be especially relevant to Mac machines, where installing several of the optional packages used by CREDO is non-trivial.
These packages are optional, they provide extra capabilities to CREDO such as plotting and visualisation of output, but are not essential for running system tests and doing basic analysis.
Currently, CREDO is primarily designed to be distributed with the stgUnderworld application bundle. It is possible to install it on its own, but currently the instructions below focus on this main use case.
To obtain CREDO as part of stgUnderworld:
e.g. use the following command:
hg clone https://www.mcc.monash.edu.au/hg/stgUnderworld
run obtainRepositories.py to download all the sub-packages, including CREDO:
- (You may have to submit your repository authentication details while cloning both repositories such as Experimental above, if you have specified to download them.)
You should now have a working version of the stgUnderworld codebase installed, including CREDO.
If you only intend to run CREDO System tests via SCons commands like ./scons.py check (see Running a test target, or test suite, via the SCons build system), then you don’t need to read the section below, as CREDO is now integrated with SCons. However, the environment variables are needed if you want to run CREDO tests directly.
To run any CREDO scripts directly, you need to modify a couple of shell environment variables.
These variables are:
|Variable||Value to set to|
|PATH||needs to be extended with a reference to the credo/scripts directory in your checkout.|
|PYTHONPATH||needs to be extended to reference the main tree of CREDO python code (credo/credo)|
|STG_BASEDIR||specifies the base directory that StGermain has been checked out to. Optional, can individually specify the variables below instead if necessary.|
|STG_BINDIR||needs to specify the path that StGermain executables have been compiled and installed to. For a default installation, you can just use STG_BASEDIR instead and CREDO will work out the binaries location within that.|
|STG_XMLDIR||needs to specify the path that StGermain standard XMLs are stored in when the code is compiled. For a default installation, you can just use STG_BASEDIR instead and CREDO will work out the XMLs location within that.|
There are also several environment variables specific to different ‘job launchers’ that you can use with CREDO, starting with the default MPI one:
MPI Jobrunner (default)
|Variable||Value to set to|
|MPI_RUN_COMMAND||(Optional) this will set the command used to run parallel jobs using MPI. Otherwise the default of “mpirun” will be used.|
The sections below will advise you how to set these up correctly.
If you would like to manually set up these environment variables, just first work out the correct values, and set them in your shell. E.g. if your stgUnderworld checkout with CREDO included was located at ~/AuScopeCodes/stgUnderworldE, then in Bash you would type:
export PATH=$PATH:~/AuScopeCodes/stgUnderworldE/credo/scripts/ export PYTHONPATH=$PYTHONPATH:~/AuScopeCodes/stgUnderworldE/credo/credo/ export STG_BINDIR=~/AuScopeCodes/stgUnderworldE/build/bin/
You might like to then save these lines to a config file for when you log in.
Alternatively, a Bash script that does all the necessary exports once you specify one single path, has been included as updatePathsCREDO.sh in the base directory of the stgUnderworld repository.
So you can just source this file into your environment each time you want to start a session and use CREDO:
you will then be ready to use CREDO.
It’s easy to test if these environment variables have been set up correctly - just open a Python script and test that you can import CREDO:
psunter@auscope-02:~/AuScopeCodes/stgUnderworldE$ python Python 2.6.4 (r264:75706, Dec 7 2009, 18:43:55) [GCC 4.4.1] on linux2 Type "help", "copyright", "credits" or "license" for more information. >>> import credo >>>
No message is the expected result, it means the credo package was successfully loaded.
If there’s an error, you will see something like:
[GCC 4.4.1] on linux2 Type "help", "copyright", "credits" or "license" for more information. >>> import credo Traceback (most recent call last): File "<stdin>", line 1, in <module> ImportError: No module named credo >>>
...which means you need to go back through the steps - most likely it’s a problem with the setup of the environment variables above.