Previously we just ported over the simple "string", "list", and "map" type
hint keywords from the old loader, which exist primarily as hints to the
CLI for whether to treat -var=... arguments and environment variables as
literal strings or as HCL expressions.
However, we've been requested before to allow more specific constraints
here because it's generally better UX for a type error to be detected
within an expression in a calling "module" block rather than at some point
deep inside a third-party module.
To allow for more specific constraints, here we use the type constraint
expression syntax defined as an extension within HCL, which uses the
variable and function call syntaxes to represent types rather than values,
like this:
- string
- number
- bool
- list(string)
- list(any)
- list(map(string))
- object({id=string,name=string})
In native HCL syntax this looks like:
variable "foo" {
type = map(string)
}
In JSON, this looks like:
{
"variable": {
"foo": {
"type": "map(string)"
}
}
}
The selection of literal processing or HCL parsing of CLI-set values is
now explicit in the model and separate from the type, though it's still
derived from the type constraint and thus not directly controllable in
configuration.
Since this syntax is more complex than the keywords that replaced it, for
now the simpler keywords are still supported and "list" and "map" are
interpreted as list(any) and map(any) respectively, mimicking how they
were interpreted by Terraform 0.11 and earlier. For the time being our
documentation should continue to recommend these shorthand versions until
we gain more experience with the more-specific type constraints; most
users should just make use of the additional primitive type constraints
this enables: bool and number.
As a result of these more-complete type constraints, we can now type-check
the default value at config load time, which has the nice side-effect of
allowing us to produce a tailored error message if an override file
produces an invalid situation; previously the result was rather confusing
because the error message referred to the original definition of the
variable and not the overridden parts.
Although we do still consider these deprecated for 0.12, we'll defer
actually generating warnings for them until a later minor release so that
module authors can retain their quoted identifiers for a period after 0.12
release for backward-compatibility with Terraform 0.11.
The initial pass of implementation here missed the special case where
ignore_changes can, in the old parser, be set to ["*"] to ignore changes
to all attributes.
Since that syntax is awkward and non-obvious, our new decoder will instead
expect ignore_changes = all, using HCL2's capability to interpret an
expression as a literal keyword. For compatibility with old configurations
we will still accept the ["*"] form but emit a deprecation warning to
encourage moving to the new form.
In our new loader we are changing certain values in configuration to be
naked keywords or references rather than quoted strings as before. Since
many of these have been shown in books, tutorials, and our own
documentation we will make the old forms generate deprecation warnings
rather than errors so that newcomers starting from older documentation
can be eased into the new syntax, rather than getting blocked.
This will also avoid creating a hard compatibility wall for reusable
modules that are already published, allowing them to still be used in
spite of these warnings and then fixed when the maintainer is able.
Much like TestParserLoadConfigFileSuccess, this is intended to be an
easy-to-maintain collection of bad examples to test different permutations
of our error handling.
As with TestParserLoadConfigFileSuccess, we should also have more specific
tests alongside this that check that the error outcome is what was
expected, since this test just accepts any error and may thus not be
testing what we think it is.
This test is intended to be an easy-to-maintain catalog of good examples
that we can use to catch certain parsing or decoding regressions easily.
It's not a fully-comprehensive test since it doesn't check the result
of decoding, instead just accepting any decode that completes without
errors. However, an easy-to-maintain test like this is a good complement
to some more specialized tests since we can easily collect good examples
over time and just add them in here.
Martin Atkins