This is true across environments, but it’s very confusing that you need to duplicate all of the values from parent in branched config. UI looks like it’s branching. In git conceptually it’s not loosing data from parent. Currently:
test
|> test_e2e
|> test_ci
|> test_unit
|> test_perf
If I have the same NODE_ENV=test in test, I expect it to be in every one of the children. And I can override it in specific, like test_perf NODE_ENV=production. In practice I have dozens of variables, where I set them in the root and expect them to propagate, and deploy explodes with missing variable, so I go and copy it to 4 children. The only edge-case here is where I want to actually drop variable, but I’d rather mark it as “absent” similar to empty value, so things like: if (isUndefined(process.env.ANALYZE)) will still work.
