bibxml-service
bibxml-service copied to clipboard
Further slim down the image
We could further slim down the production image by excluding a lot of things: for example, things we needed to build static (Node, etc.) and things we needed to build back-end dependencies (and even git
is only required by celery
service).
Typically this seems to be done via intermediate images, things are build there and then only the results are copied into the next image.
It looks like getting rid of Node could be easy this way, because build results are directly under /code/…
. Getting rid of other stuff is trickier, I think built Python dependencies are outside of /code
(probably in site-packages
somewhere).
Complicating factor is that rebuilding e.g. static is required during development, so we’d need to maintain a separate image that keeps Node in.