Multi-software projects#

Many projects will use multiple softwares, along the lines of “use what is appropriate for specific tasks.” This may pose particular challenges. I showed that a generic bash script can handle running multiple softwares in the right sequence, but some researchers may be more comfortable in a main software package, or may not want to incur the additional complication of relying on bash, or worse, restricting reproducibility to a particular platform, for instance by choosing cmd.exe (.bat files) or Powershell scripts.

System Paths#

In order to call external software from within a statistical programming language, it is necessary to know where, on a system, the software can be found. In “Environments”, we discussed search paths, but in many computing environments, not each piece of software is on that search path.

However, it is possible to explicitly address this. Some software is aware of its own installation path.

R.home("bin")

will yield the location of the R executable, for instance C:/Program Files/R/R-4.4.0/bin/x64/R. Note that it will vary by version!

display c(sysdir_stata)

might display

. display c(sysdir_stata)
/usr/local/stata/

Others may require querying the operating system itself:

For instance, to find Python, typing

which python3

into a Bash or Zsh window might yield /usr/bin/python3 on a particular *nix system (including macOS).

(Get-Command python).Path

into a (Windows) Powershell window might yield C:\Program Files\Python312\python.exe on a particular Windows system.

where python

into a (Windows) Command window might yield C:\Program Files\Python312\python.exe on a particular Windows system.

Calling software A from within software B#

Calling other software from Stata#

The path information can then be used, for instance, to call R and python from a Stata program:

global Rterm_path "C:/Program Files/R/R-4.3.1/bin/x64"
* Set path to python executable here:
global python_ex "C:\Program Files\Python312\python.exe"

ssc install rsource

python set exec "$python_ex"
python script "code/run_geo_clean.py"

rsource using "code/figure1.R"

global Rterm_path "C:/Program Files/R/R-4.3.1/bin/x64"
* Set path to python executable here:
global python_ex "C:/Program Files/Python312/python.exe"

ssc install rsource
ssc install python


set python_exec "$python_ex"
python script "code/run_geo_clean.py"

rsource using "code/figure1.R"

where the Stata package rsource is used.[1]

Calling other software from Python#

(WIP)

Python can be used as a multi-purpose orchestration system quite easily. Multiple “kernels” can be used in Jupyter notebooks, with appropriate support. Very simply, any OS call from Python can initiate “command line” execution of other code. More involved is to have programmatic structured approach. Examples include:

  • pystata: calling Stata from Python