[WIP]: provider/aws: Add aws_availability_zones data source

[jen20]

This PR adds a data source with a single set, available for the schema which gets populated with the availability zones to which an account has access.

I implemented it as part of the review process of data sources so thought I would open an additional PR for feedback from @apparentlymart about this implementation.

One area we need to consider is nomenclature (cc @phinze, @catsby, @mitchellh) - many of these sources will return lists, and we should take care to keep naming arity consistent - I am expecting the actual names here to change when we make that decision.

TODO: Acceptance test.

[vancluever]

@jen20 you might want to check #4848 as well, that one sorts the AZs (to make sure the order is predictable) and there's an acceptance test there too.

[apparentlymart]

@jen20 my thought was that to start we'd mainly focus on data sources that return single objects... so aws_ami returns a single AMI matching given criteria, etc... because right now you can't really do much with attributes that are lists of resources in Terraform anyway.

However, this one I think makes sense because ultimately it's just a list of strings, which Terraform is pretty good at dealing with. And so calling it aws_availability_zones makes sense.

I was planning to build out aws_ami (using @vancluever's earlier work as a starting point) as an initial non-trivial data source example, establishing a pattern for how AWS Describe* actions can map on to data sources... but I was planning to wait until #6598 lands before building that, in case we make any changes to the interface before it is merged.

[apparentlymart]

Does this second underscore also need to be escaped, or will Markdown ignore it because there's an odd number of them? (In other files I've seen all the underscores escaped and have been following that pattern, but not sure if that's actually necessary...)

[jen20]

Probably. I hate markdown ;-)

[apparentlymart]

Looks good to me, minor inline niggles aside. 🎺

[apparentlymart]

Certainly doesn't have to be in the first version of this data source, but I was thinking that when mapping Describe* calls onto data sources we'd map each of the filters onto an Optional attribute and then include them in the call.

In this case, that could be attributes regions, which would be a list of regions to constrain to, and states, allowing to filter by state.

So then I'd use it something like this:

data "aws_availability_zones" "us" {
    regions = ["us-west-1", "us-west-2", "us-east-1"]
    states = ["available", "information", "impaired"] # don't include "unavailable" AZs
}

[jen20]

@apparentlymart I'm not sure that I see the value in this case, since the configured provider will determine which region is valid for use? In general though, I agree with this.

[jen20]

I have no idea why the diff is so screwed up here - I thought I had rebased onto origin/master. I'll close this one and open a new PR for this soon.

[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.