The current pyinstaller pattern is not going to work. Trying to follow similar patterns such as make self will not result in a useful portable pyinstaller executable, because it can rely on the python in a docker image, which itself, will not be portable.
Using a dedicated default image for pyinstaller, will not work, because there is too much potential for it to be incompatible with the running image for a project, so this will not solve anything else.
The correct solution would be to add docker command to the end of a docker image, to say, install a potable version of python (using anaconda again). This should be relatively straight forward, using
ARG OS
FROM ${OS}
... check python version, install miniconda closest to it
... setup pyinstaller
The current pyinstaller pattern is not going to work. Trying to follow similar patterns such as make self will not result in a useful portable pyinstaller executable, because it can rely on the python in a docker image, which itself, will not be portable.
Using a dedicated default image for pyinstaller, will not work, because there is too much potential for it to be incompatible with the running image for a project, so this will not solve anything else.
The correct solution would be to add docker command to the end of a docker image, to say, install a potable version of python (using anaconda again). This should be relatively straight forward, using