You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Is your feature request related to a problem? Please describe.
Current installation of the Atlas clients (the cli, the SDK...all the stuff on the user side) is tied to a local installation of server-side software, unless the right flags are set during installation. There is a more widely understood way to install these clients via pip and PyPI that we can take advantage of to reduce friction for users and simplify installation.
For context, we did have a private repo such that our internal users and developers could install the clients via pip. We didn't create packages on PyPI at that time (when Atlas was still closed-source) because we didn't want to expose the source code so widely. But now, this is no longer a concern.
Describe the solution you'd like
Get these clients into one/several PyPI packages via CD
Modify documentation on website, github, the official docs, and anywhere else to have a client-only installation guide that is essentially a one-liner pip install xxx yyy
Modify other parts of documentation that can take advantage of this feature: for example, custom docker worker images that want to use the Atlas clients will be much easier to create by just having the pip install line in either the dockerfile or in an entrypoint script.
Modify the installer to install these clients using pip. (Optionally) Change up the installer UI for a more sensible default behaviour and flow, separating out clients installations and server installation.
Things worth discussing
How many PyPI packages and what should they be called?
Clearly delineate what's client and server and potentially clean up some terminology (e.g. someone who doesn't use the SDK or the CLI, but uses the GUI is still a "user", but won't need to install anything)
The text was updated successfully, but these errors were encountered:
Is your feature request related to a problem? Please describe.
Current installation of the Atlas clients (the cli, the SDK...all the stuff on the user side) is tied to a local installation of server-side software, unless the right flags are set during installation. There is a more widely understood way to install these clients via
pip
and PyPI that we can take advantage of to reduce friction for users and simplify installation.For context, we did have a private repo such that our internal users and developers could install the clients via
pip
. We didn't create packages on PyPI at that time (when Atlas was still closed-source) because we didn't want to expose the source code so widely. But now, this is no longer a concern.Describe the solution you'd like
pip install xxx yyy
pip install
line in either the dockerfile or in an entrypoint script.pip
. (Optionally) Change up the installer UI for a more sensible default behaviour and flow, separating out clients installations and server installation.Things worth discussing
The text was updated successfully, but these errors were encountered: