3 ms·
Efficient management Python projects dependencies with Docker
- dozzie 10y agoI like how Docker crowd seems to be undereducated about what packaging systems can do for dozen years already. - If something not packaged by distribution is needed, one can just create own package. - If something is packaged, but in wrong version, most of the time the package can be simply rebuilt. - If a required library is fscked up, doesn't cooperate and needs to be put into a virtualenv, whole virtualenv still may be put into a package. Installation and deployment is then a breeze, pulling all the dependencies (including C libraries), and not compiling everything over and over again. Moreover, deployment is repeatable, which means a developer may install project's dependencies and work on the code directly from git workdir. A side bonus is immunity from another instance of left-pad farce, where inaccessibility of one single tarball hanging somewhere in the internet broke down whole build process.