We're now writing the "planned new value" to OutputValue, but the data
resource nodes during refresh need to see the verbatim config value in
order to decide whether read must be deferred to the apply phase, so we'll
optionally export that here too.
Our state representation is not able to preserve unknown values, so it's
not suitable for retaining the transient incomplete values we produce
during planning.
Instead, we'll discard the unknown values when writing to state and have
the expression evaluator prefer an object from the plan where possible.
We still use the shape of the transient state to inform things like the
resource's "each mode", so the plan only masks the object values
themselves.
This is no longer a call into the provider, since all of the data diff
logic is standard for all data sources anyway. Instead, we just compute
the planned new value and construct a planned change from that as-is.
Previously the provider could, in principle, customize the read diff. In
practice there is no real reason to do that and the existing SDK didn't
pass that possibility through to provider code, so we can safely change
this without impacting provider compatibility.
Previously we just left these out of the plan altogether, but in the new
plan types we intentionally include change information for every resource
instance, even if no changes are actually planned, to allow alternative
plan file viewers to show what isn't changing as well as what is.
This also includes passing in the provider schema to a few more EvalNodes
that were expecting it but not getting it, in order to be able to
successfully test the implementation of EvalReadDiff here.
This codepath is going to be significantly changed before release to make
it support structural diff of the new data types, but this lets us lean on
the old renderer to produce partial output in the mean time while we
continue to work on getting things working end-to-end after the
considerable refactoring that's been going on.
Previously we would construct both provisioner and the provider objects if
either callback was set, but this is incorrect because a plugin should
actually set only one of these at a time, depending on what kind of plugin
it is.
This includes the new PathSet type, which we'll use to represent the
"requires replacement" set of attribute paths coming back from providers
during planning.
This includes a bugfix to the cty/msgpack package to ensure correct
decoding of unknown and null values.
This also includes updates to cty's dependencies.
govendor's lack of understanding of transitive versions was making it hell
to figure out some dependency hell between etcd, grpc, and protobuf. After
fighting with it for a few hours, I decided to give Go 1.11rc2 a try since
previous experiments had been promising on the 1.11 tree in master.
The dependencies all worked out first time when managed using the Go
Modules code, and so we'll run with this now to continue to make progress
though we may wish to back out of this nearer to release and return to
govendor for a while until other projects have caught up.
However, since this commit includes a vendor directory produced using Go
Modules it doesn't actually _require_ Go 1.11 to build, and instead
requires it only to make further changes to the selected versions in the
vendor dir. Go 1.10's vendoring support will still find the modules in
their expected locations within the vendor dir.
We need to make the collection itself be a tuple or object rather than
list or map in this case, since otherwise all of the elements of the
collection are constrained to be of the same type and that isn't the
intent of a provider indicating that it accepts any type.
This function's goal is to ensure that the "final" plan value produced
by a provider during the apply step is always consistent with the known
parts of the planned value produced during the plan step.
Any error produced here indicates a bug in the provider.
This produces a "proposed new state", which already has prior computed
values propagated into it (since that behavior is standard for all
resource types) but could be customized further by the provider to make
the "_planned_ new state".
In the process of implementing this it became clear that our configschema
DecoderSpec behavior is incorrect, since it's producing list values for
NestingList and map values for NestingMap. While that seems like it should
be right, we should actually be using tuple and object types respectively
to allow each block to have a different runtime type in situations where
an attribute is given the type cty.DynamicPseudoType. That's not fixed
here, and so without a further fix list and map blocks will panic here.
The DecoderSpec implementation will be fixed in a subsequent commit.
This allows a provider to retain arbitrary extra data in the plan and
make use of it during apply. The contents are not used by Terraform and
never shown to the user.
We're going to allow the provider to encode whatever it wants in here, so
a provider can use whatever is most convenient for its implementation
language and to avoid some of the bugs we saw with the prior model where
the forced round-trip through JSON and back into interface{} would cause
some loss of fidelity, leading to bugs.
Chaange ResourceProvider to providers.Interface starting from the
context, and fix all type errors.
This only replaced some of method calls directly applicable to the
providers themselves. The resource methods will follow.
This is a general catchup of some developments in the HCL2 codebase, but
in particular includes:
- Recording expression and evalcontext as part of diagnostics, so that
variable value information can be included alongside diagnostic
snippets.
- hcldec supports decoding blocks into tuple and object values as well as
list and map values, which then allows cty.DynamicPseudoType nested
attributes to work properly.
These new types are adaptations of terraform.ProviderResolver and
terraform.ResourceProviderFactory respectively. In a later commit we will
update users of the old types to use these new types, possibly in a
"gradual repair" sort of fashion by initially using type aliases, though
that won't 100% solve the problem because the types themselves have
changed in this move to return providers.Interface instead of
terraform.ResourceProvider.
MockProvider and MockProvisioner implement the new plugin interfaces,
and are built following the patterns used by the legacy
MockResourceProvider and MockResourceProvisioner
Close was previously an optional interface, because it's only applicable
on the client side but the types were shared accross all packages. Since
all plugins will now implement Close, it no longer needs to be optional.
Managing which function need to be shared between the terraform plugin
and the helper plugin without creating cycles was becoming difficult.
Move all functions related to converting between terraform and proto
type into plugin/convert.
Martin Atkins