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
Eventually, I'd like to move our base image into this project also. I think scheduled rebuilds are a worthwhile concept, to ensure that stuff gets kept up-to-date.
Once we split out our various components into their own repositories, it will be semi-trivial to think about dockerizing each one of them as a service, thus making things a bit easier for connecting and turning into a clustered service.
Here is what I am thinking would be nice. We should synchronize all versions together in this repository. (That is, this repository would be the master for all of our dependency updates.) New versions of clustal, new versions of everything, would require actual code changes, but we would not tag anything here. (No need to.)
Only new commits on master would get deployed.
New commits on separate branches would get built and tagged, but not deployed.
Code should only get submoduled here if possible; if substantial changes are made to a component, then those changes should go to a different repository with a separate build cycle.
@hawkrives, does this sound alright? This way, we have a canonical representation of what our images are, and it's all in one place. Version updates are as simple as just changing a number in a file, then.
The above image kinda sketches out my ideas for how this would get structured. The CI build would be structured the same, so we would ensure that all new base image build changes are reflected throughout the entire image cascade.
The text was updated successfully, but these errors were encountered:
Eventually, I'd like to move our base image into this project also. I think scheduled rebuilds are a worthwhile concept, to ensure that stuff gets kept up-to-date.
Once we split out our various components into their own repositories, it will be semi-trivial to think about dockerizing each one of them as a service, thus making things a bit easier for connecting and turning into a clustered service.
Here is what I am thinking would be nice. We should synchronize all versions together in this repository. (That is, this repository would be the master for all of our dependency updates.) New versions of clustal, new versions of everything, would require actual code changes, but we would not tag anything here. (No need to.)
Only new commits on master would get deployed.
Code should only get submoduled here if possible; if substantial changes are made to a component, then those changes should go to a different repository with a separate build cycle.
@hawkrives, does this sound alright? This way, we have a canonical representation of what our images are, and it's all in one place. Version updates are as simple as just changing a number in a file, then.
The above image kinda sketches out my ideas for how this would get structured. The CI build would be structured the same, so we would ensure that all new base image build changes are reflected throughout the entire image cascade.
The text was updated successfully, but these errors were encountered: