The graph walking mechanism is specified as requiring a graph with a single
root, which in practice means there's exactly one node in the graph
which doesn't have any dependencies.
However, we previously weren't verifying that invariant is true for
subgraphs returned from DynamicExpand. It was working anyway, but it's not
ideal to be relying on a behavior that isn't guaranteed by our underlying
infrastructure.
We also previously had the RootTransformer being a bit clever and trying
to avoid adding a new node if there is already only a single graph with
no dependencies. That special case isn't particularly valuable since
there's no harm in turning a one-node graph into a two-node graph with
an explicit separate root node, and doing that allows us to assume that
the root node is always present and is always exactly terraform.rootNode.
Many existing DynamicExpand implementations were not producing valid
graphs and were previously getting away with it. All of them now produce
properly-rooted graphs that should pass validation, and we will guarantee
that with an explicit check of the DynamicExpand return value before we
try to walk that subgraph. For good measure we also verify that the root
node is exactly terraform.rootNode, even though that isn't strictly
required by our graph walker, just to help us catch potential future bugs
where a DynamicExpand implementation neglects to add our singleton root
node.
We were previously _trying_ to handle diagnostics here but were not quite
doing it right because we were testing whether the resulting error was
nil rather than appending it to the diagnostics and then seeing if the
result has errors.
The difference here is important because it allows DynamicExpand to return
warnings without associated errors when needed. Previously the graph
walker would treat a warnings-only result as if it were an error.
Ideally we'd change DynamicExpand to return diagnostics directly, but we
previously decided against that because there were so many implementors
to update, and my intent for this change is to be surgical in the update
so we minimize risk of backporting the change into patch releases.
Previously terraform.Context was built in an unfortunate way where all of
the data was provided up front in terraform.NewContext and then mutated
directly by subsequent operations. That made the data flow hard to follow,
commonly leading to bugs, and also meant that we were forced to take
various actions too early in terraform.NewContext, rather than waiting
until a more appropriate time during an operation.
This (enormous) commit changes terraform.Context so that its fields are
broadly just unchanging data about the execution context (current
workspace name, available plugins, etc) whereas the main data Terraform
works with arrives via individual method arguments and is returned in
return values.
Specifically, this means that terraform.Context no longer "has-a" config,
state, and "planned changes", instead holding on to those only temporarily
during an operation. The caller is responsible for propagating the outcome
of one step into the next step so that the data flow between operations is
actually visible.
However, since that's a change to the main entry points in the "terraform"
package, this commit also touches every file in the codebase which
interacted with those APIs. Most of the noise here is in updating tests
to take the same actions using the new API style, but this also affects
the main-code callers in the backends and in the command package.
My goal here was to refactor without changing observable behavior, but in
practice there are a couple externally-visible behavior variations here
that seemed okay in service of the broader goal:
- The "terraform graph" command is no longer hooked directly into the
core graph builders, because that's no longer part of the public API.
However, I did include a couple new Context functions whose contract
is to produce a UI-oriented graph, and _for now_ those continue to
return the physical graph we use for those operations. There's no
exported API for generating the "validate" and "eval" graphs, because
neither is particularly interesting in its own right, and so
"terraform graph" no longer supports those graph types.
- terraform.NewContext no longer has the responsibility for collecting
all of the provider schemas up front. Instead, we wait until we need
them. However, that means that some of our error messages now have a
slightly different shape due to unwinding through a differently-shaped
call stack. As of this commit we also end up reloading the schemas
multiple times in some cases, which is functionally acceptable but
likely represents a performance regression. I intend to rework this to
use caching, but I'm saving that for a later commit because this one is
big enough already.
The proximal reason for this change is to resolve the chicken/egg problem
whereby there was previously no single point where we could apply "moved"
statements to the previous run state before creating a plan. With this
change in place, we can now do that as part of Context.Plan, prior to
forking the input state into the three separate state artifacts we use
during planning.
However, this is at least the third project in a row where the previous
API design led to piling more functionality into terraform.NewContext and
then working around the incorrect order of operations that produces, so
I intend that by paying the cost/risk of this large diff now we can in
turn reduce the cost/risk of future projects that relate to our main
workflow actions.
This is part of a general effort to move all of Terraform's non-library
package surface under internal in order to reinforce that these are for
internal use within Terraform only.
If you were previously importing packages under this prefix into an
external codebase, you could pin to an earlier release tag as an interim
solution until you've make a plan to achieve the same functionality some
other way.
