As of today, the only way to publish a shared library is via a git repository. The possibility to reference a tag is already a decent way to get a fixed version of the library. But having to clone a git repository is probably not the most convenient way to retrieve some code / libraries. I would have felt more comfortable if the library retrieval had been done via an http request for example.
I see two advantages there:
- a faster download (I don't need any commit comment, changes history ... or any info that git downloads on top of the code itself)
- the download would rely (at least in our case) on services with different SLAs compared to our git repository. This is even more meaningful for Pipelines that are not triggered by SCM changes. By having the code of the shared library packaged as tar.gz on Artifactory (in our case) or anything available via http/s, we would avoid the dependency on git