Stack/project organizational best-practices question. Say I have a several-to-many situation I want to organize. I have multiple

foo

resources (auth providers) and multiple

bar

stacks (application environments). Each

bar

needs to talk to one and only one

foo

, and thus needs access to

foo

exported values. Each

foo

may have multiple `bar`s talking to it, but doesn't need to know anything about them. The `bar`s exist today,

foo

is new. How would folks organize this?

foo

-specific project? My ideal would be when creating a new

bar

that does not yet have an associated

foo

pulumi would magically create it, however I also don't want to risk the first

foo

being deleted and taking the

bar

with it unless all the `bar`s using

foo

were also gone.

There are two things you need to consider about • change ratio of resource • and resource lifecycle I can understand you don’t want to delete all bars unless all the bars using foo were also gone, actually I don’t think we should imitate infrastructure components as behaviorally with software components, let’s say, you can’t( actually you can but you gonna add just too much complexity into your iac codebase) use a bucket algorithm( this is just for example) so if bucket is empty then delete all bars, if it is not empty then it means there are foo. So I’d rather go one-to-one and ship the all foo with their own bars

Ah. I clearly left out that a) we cannot have a 1:1 in this case b) these are all super long lived, on the order of years. We'll update them all periodically, some frequently, but they're more pets than cattle.