[WIP] helper/schema: implement ComputedWhen #4846
Conversation
| // finished the diff. | ||
| // TODO | ||
| for k, schema := range m { |
There was a problem hiding this comment.
Here be dragons. Lots of looping logic below. Would love to come up with a more elegant way to do this. Considered attempting to thread ComputedWhen knowledge/context into the original diff generation rather than this post-hoc munging, but initial spikes felt even more messy than this.
ComputedWhen is a long `TODO`d schema feature that allows any field to be marked as computed when some terget field changes. Right now Terraform has no way of invalidating a field until it sees _that_ field changes, but there are plenty of examples of scenarios where one field changing implies that another field is going to be assigned a new value.
phinze
force-pushed
the
phinze/computed-when
branch
from
0914559
to
2761e3c
Compare
|
ComputedWhen is still a good idea to implement, as it has the potential to trigger richer plan-time output, but this is an old PR and I have no plans to pick it up in the foreseeable future. /cc @jbardin to put this on your "someday" radar 💌 |
|
@jbardin: is there any update on this issue of the implementation of |
|
Hi @mavogel, I think this use case is now more generally covered by the new |
|
I'm going to lock this issue because it has been closed for 30 days ⏳. This helps our maintainers find and focus on the active issues. If you have found a problem that seems similar to this, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further. |
ghost
locked and limited conversation to collaborators
ComputedWhen is a long
TODOd schema feature that allows any field tobe marked as computed when some target field changes.
Right now Terraform has no way of invalidating a field until it sees
that field changes, but there are plenty of examples of scenarios
where one field changing implies that another field is going to be
assigned a new value.
I'd like to do a bit more testing and refactoring before this is fully ready. But I wanted to push what I had for now since it seems to be working for the cases I've tried it.
This should improve the experience of using consul_keys and atlas_artifacts (for scenarios where two plan/apply cycles would be required for a change to propagate), as well as any other resources where we find similar scenarios that we'd like to use
ComputedWhen.