Most of the NCSSAR home-brew python repo's don't use version numbering. While the clean way to do things would be to add version numbering / tag numbering to each repo, that's not completely reliable - what if a hotfix is made but no new tag is or the version number is not incremented? Then a simple version# / tag# comparison would not show a need for an update.
So, to be safer, compare the contents of the installation to the contents of the repo head. A recursive diff of the unzipped downloaded repo head would be one way.
To avoid doing this download-and-compare more often than necessary, maybe keep track of when the latest comparison was run, and what the file modification times were when the comparison was run, and run a comparison again if there are any file modification times more recent than the last comparison? Would need to account for config files / 'local' directories inside the unzipped dir tree, which would intentionally differ from the repo head.
Is there a better way?