make diagnose on for_each argument more precise

[tmatias]

That condition would not diagnose other iterables, such as tuple argument values, causing a panic: not a string later on.

[hashicorp-cla]

All committers have signed the CLA.

[teamterraform]

Hi @tmatias, thanks for your contribution!

It looks like this change causes the TestContext2Plan_forEach test to fail - please take a look at the Travis build log for details.

[pselle]

Thanks @tmatias!

[omeid]

This doesn't seem right. Shouldn't you be able to use list type with for_each?

cc: @teamterraform

[teamterraform]

@omeid resources created with for_each are identified by their string key, and as such, values must be unique. It's for that reason that Terraform enforces using sets rather than lists of strings, because it should not be ambiguous what resources will be created.

[omeid]

In that case then the documentation needs to be updated accordingly, from the intro to the actual configuration document suggests that lists should work.

Intro: ...includes the idea of iterating over a list or map using the for_each argument, which is intended as a more.

expressions docs:
Since the for_each argument accepts any collection or structural value.

Other than the documentation, I think the general that for_each = resource.multiple.*.id should work is a very pragmatic expectation.

[ghost]

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.