There’s a breaking change coming to Baremetal in the way that code is updated. Currently it’s a git pull, but going forward it’s a git clone. The breaking part is that you’ll have to go to the server and recreate the directory structure (see the PR for an exhaustive description).
For those using Baremetal: is this going to ruin your life or is this something you can deal with? If you have a production application this will definitely incur downtime, unless you went down the path of starting up a separate pool of servers from scratch (while leaving the existing ones running) and then swapping over load balancer members or switching over DNS. This code update strategy is much more robust that the original pull version and there aren’t any plans to change it once this PR is merged, so this should be the only time you’ll need to make a major change manually on the server going forward.
This PR (and several others that add stuff like multiple environment support, clean up old releases, rollbacks) will most likely be merged in version 2.0 of Redwood, although we don’t currently have a release date in mind.
We threw around the idea of marking Baremetal as “experimental” and thus not subject to semantic versioning rules (a major revision is required to make a non-backwards compatible change), but that doesn’t seem fair since it was launched in 1.0 without the “experimental” label.
Let us know if you have any concerns before we merge it!