These changes restore a more-correct version of the behavior that was
disabled with #3014. The original implementation of this behavior was
done in the SDKs, which do not have access to the complete inputs for a
resource (in particular, default values filled in by the provider during
`Check` are not exposed to the SDK). This lack of information meant that
the resolved output values could disagree with the typings present in
a provider SDK. Exacerbating this problem was the fact that unknown
values were dropped entirely, causing `undefined` values to appear in
unexpected places.
By doing this in the engine and allowing unknown values to be
represented in a first-class manner in the SDK, we can attack both of
these issues.
Although this behavior is not _strictly_ consistent with respect to the
resource model--in an update, a resource's output properties will come
from its provider and may differ from its input properties--this
behavior was present in the product for a fairly long time without
significant issues. In the future, we may be able to improve the
accuracy of resource outputs during a preview by allowing the provider
to dry-run CRUD operations and return partially-known values where
possible.
These changes also introduce new APIs in the Node and Python SDKs
that work with unknown values in a first-class fashion:
- A new parameter to the `apply` function that indicates that the
callback should be run even if the result of the apply contains
unknown values
- `containsUnknowns` and `isUnknown`, which return true if a value
either contains nested unknown values or is exactly an unknown value
- The `Unknown` type, which represents unknown values
The primary use case for these APIs is to allow nested, properties with
known values to be accessed via the lifted property accessor even when
the containing property is not fully know. A common example of this
pattern is the `metadata.name` property of a Kubernetes `Namespace`
object: while other properties of the `metadata` bag may be unknown,
`name` is often known. These APIs allow `ns.metadata.name` to return a
known value in this case.
In order to avoid exposing downlevel SDKs to unknown values--a change
which could break user code by exposing it to unexpected values--a
language SDK must indicate whether or not it supports first-class
unknown values as part of each `RegisterResourceRequest`.
These changes also allow us to avoid breaking user code with the new
behavior introduced by the prior commit.
Fixes#3190.
- Do not use a non-zero-to-zero transition in the number of outstanding
RPCs to determine the completion of a Python program until after the
synchronous piece of the program has finished running is complete
- Instead of using a future to indicate that either a) a zero-to-one
transition in the number of outstanding RPCs has occurred, or b) an
unhandled exception has occurred, a) observe the transition itself,
and b) use an optional exception field to track the presence or
absence of an exception.
Fixes#3162.
_sync_await was not reentrant with respect to _run_once: the latter
captures the length of the ready list before it iterates it, and the
former drains the ready list by reentering _run_once. Fix this by
tracking the length of the list before pumping the event loop and then
pushing cancelled handles on to the list as necessary after pumping the
loop.
These changes also fix an issue with `export`ing awaitables.
Fixes#3038.
These changes fix a bug in the Python runtime that would cause any
awaitable input properties passed to a resource that are missing
from the resource's output properties to be awaited twice. The fix is
straightforward: rather than roundtripping an input property through
serialize/deserialize, just deserialized the already-serialized input
property.
Fixes#2940.
These changes make the `pulumi.runtime.invoke` function invokable in a
synchronous manner. Because this function still needs to perform
asynchronous work under the covers--namely awaiting a provider URN and
ID if a provider instance is present in the `InvokeOptions`--this
requires some creativity. This creativity comes in the form of a helper
function, `_sync_await`, that performs a logical yield from the
currently running event, manually runs the event loop until the given
future completes, performs a logical resume back to the
currently executing event, and returns the result of the future.
The code in `_sync_await` is a bit scary, as it relies upon knowledge of
(and functions in) the internals of the `asyncio` package. The necessary
work performed in this function was derived from the implementations of
`task_step` (which pointed out the need to call `_{enter,leave}_task`)
and `BaseEventLoop.run_forever` (which illustrated how the event loop is
pumped). In addition to potential breaking changes to these internals,
the code may not work if a user has provided an alternative implementation
for `EventLoop`. That said, the code is a close enough copy of
`BaseEventLoop.run_forever` that it should be a reasonable solution.
* Remove pulumi.io reference in tests
* Remove pulumi.io references in Dockerfiles
* Remove pulumi.io references in CONTRIBUTING.md
* Update README.md's
* Use correct logo
A resource can be imported by setting the `import` property in the
resource options bag when instantiating a resource. In order to
successfully import a resource, its desired configuration (i.e. its
inputs) must not differ from its actual configuration (i.e. its state)
as calculated by the resource's provider.
There are a few interesting state transitions hiding here when importing
a resource:
1. No prior resource exists in the checkpoint file. In this case, the
resource is simply imported.
2. An external resource exists in the checkpoint file. In this case, the
resource is imported and the old external state is discarded.
3. A non-external resource exists in the checkpoint file and its ID is
different from the ID to import. In this case, the new resource is
imported and the old resource is deleted.
4. A non-external resource exists in the checkpoint file, but the ID is
the same as the ID to import. In this case, the import ID is ignored
and the resource is treated as it would be in all cases except for
changes that would replace the resource. In that case, the step
generator issues an error that indicates that the import ID should be
removed: were we to move forward with the replace, the new state of
the stack would fall under case (3), which is almost certainly not
what the user intends.
Fixes#1662.
There current RPC model for Pulumi allows secret values to be deeply
embedded in lists or maps, however at the language level, since we
track secrets via `Output<T>` we need to ensure that during
deserialization, if a list or a map contains a secret, we need to
instead treat it as if the entire list or map was a secret.
We have logic in the language runtimes to do this as part of
serialization. There were a few issues this commit addresses:
- We were not promoting secretness across arrays in either Node or
Python
- For Python, our promotion logic was buggy and caused it to behave in
a manner where if any value was secret, the output values of the
object would be corrupted, because we'd incorrectly treat the
outputs as a secret who's value was a map, instead of a map of
values (some of which may be secret).
This caused very confusing behavior, because it would appear that a
resource creation call just did not set various output properties when
one or more of them ended up containing a secret.
This commit implements read_resource functionality for Python in a
manner identical to the NodeJS implementation. If an "id" option is
passed to a resource via ResourceOptions on construction, that resource
will be read and not created.
* NodeJS: allow callers to override provider version
* Python: allow callers to override provider version
* NodeJS: add version for invoke
* Python: add version to invoke
* NodeJS: add tests for ReadResource
* Post-merge cleanup
* update doc comments
Fixes#2277.
Adds a new ignoreChanges resource option that allows specifying a list of property names whose values will be ignored during updates. The property values will be used for Create, but will be ignored for purposes of updates, and as a result also cannot trigger replacements.
This is a feature of the Pulumi engine, not of the resource providers, so no new logic is needed in providers to support this feature. Instead, the engine simply replaces the values of input properties in the goal state with old inputs for properties marked as ignoreChanges.
Currently, only top level properties may be specified in ignoreChanges. In the future, this could be extended to support paths to nested properties (including into array elements) with a JSONPath/JMESPath syntax.
* Enable unit testing for Pulumi programs
This change enables rudimentary unit testing of your Pulumi programs, by introducing a `PULUMI_TEST_MODE` envvar that, when set, allows programs to run without a CLI. That includes
* Just being able to import your Pulumi modules, and test ordinary functions -- which otherwise would have often accidentally triggered the "Not Running in a CLI" error message
* Being able to verify a subset of resource properties and shapes, with the caveat that outputs are not included, due to the fact that this is a perpetual "dry run" without any engine operations occurring
In principle, this also means you can attach a debugger and step through your code.
* Finish the unit testing features
This change
1) Incorporates CR feedback, namely requiring that test mode be
explicitly enabled for any of this to work.
2) Implements Python support for the same capabilities.
3) Includes tests for both JavaScript and Python SDKs.
* Add a note on unit testing to the CHANGELOG
* Use Node 8 friendly assert API
* Embellish the CHANGELOG entry a bit
* Fix an issue with empty ID for CustomResource
The Python runtime was checking the ID field it receives from the engine
against None, assuming that the engine would not set the ID field if one
was not present. However, it does set the ID field; it is set to the
empty string when an ID is not known.
This commit fixes an issue that can cause certain IDs to be erroneously
considered to be known during previews, which can cause problems during
the Check phase of resources that directly reference IDs of other
resources.
* Add CHANGELOG
These changes add a new flag to the various `ResourceOptions` types that
indicates that a resource should be deleted before it is replaced, even
if the provider does not require this behavior. The usual
delete-before-replace cascade semantics apply.
Fixes#1620.
- Add support for per-property dependencies to the Go SDK
- Add tests for first-class secret rejection in the checkpoint and RPC
layers and language SDKs
This implements the new algorithm for deciding which resources must be
deleted due to a delete-before-replace operation.
We need to compute the set of resources that may be replaced by a
change to the resource under consideration. We do this by taking the
complete set of transitive dependents on the resource under
consideration and removing any resources that would not be replaced by
changes to their dependencies. We determine whether or not a resource
may be replaced by substituting unknowns for input properties that may
change due to deletion of the resources their value depends on and
calling the resource provider's Diff method.
This is perhaps clearer when described by example. Consider the
following dependency graph:
A
__|__
B C
| _|_
D E F
In this graph, all of B, C, D, E, and F transitively depend on A. It may
be the case, however, that changes to the specific properties of any of
those resources R that would occur if a resource on the path to A were
deleted and recreated may not cause R to be replaced. For example, the
edge from B to A may be a simple dependsOn edge such that a change to
B does not actually influence any of B's input properties. In that case,
neither B nor D would need to be deleted before A could be deleted.
In order to make the above algorithm a reality, the resource monitor
interface has been updated to include a map that associates an input
property key with the list of resources that input property depends on.
Older clients of the resource monitor will leave this map empty, in
which case all input properties will be treated as depending on all
dependencies of the resource. This is probably overly conservative, but
it is less conservative than what we currently implement, and is
certainly correct.
On startup, when we were populating the Settings object, we failed to
coerce the dry_run parameter from a string to a boolean, which resulted
in is_dry_run always believing that it was a preview. This PR fixes this
oversight by explicitly coercing to a boolean prior to sending the value
to Settings.
When is resource is waiting for its dependencies to resolve, it first
informs the RPC_MANAGER that it intends to do an RPC - this is to
prevent premature termination of the program while RPCs are still in
flight or queued to execute.
However, this is a problem whenever a resource fails to create for
whatever reason. The most common ways for this to happen are for invokes
to fail, resource creation itself to fail, or throwing in an apply.
Today, this causes a deadlock, since all consumers of the failed
resources block forever while never decrementing the RPC count.
This commit addresses the issue by deliberately (abnormally) resolving
all futures that are created in the process of preparing a resource.
This solves the problem by immediately terminating all resources that
are waiting for the failed resource's outputs to resolve - they resolve
immediately, and exceptionally.
The end result is now that, instead of deadlocking, a doomed program now
terminates as expected with a single exception message.
When previewing a real first-class provider, it is often the case that
the provider's ID is unknown during a preview. This commit fixes a bug
where we did not translate an unknown ID into the rpc layer's sentinel
UNKNOWN value where we should have, which caused the engine to fail to
resolve the provider reference.
Invoke in Node.js allows users to optionally pass a parent or a provider
to the invoke, which dictates either explicitly or implicitly which
provider to use when performing an invoke. If a provider is specified
explicitly, that provider is used to perform the invoke. If a parent is
specified, that parent's provider is used to perform the invoke.
* Implement first-class providers for Python
First-class providers are an explicit projection of providers themselves
into Pulumi programs. For the most post, providers are just regular
resources, but the addition of providers to the fray (and the ability of
resources to be constructed by providers in the same program) requires
some changes to the Python resource model.
A summary of the changes:
1. Added ProviderResource, a custom resource that is the base class of
all providers.
2. ResourceOptions now has 'provider' and 'providers' fields.
'provider', when passed to a custom resource, allows users to override
the provider that is used to construct a resource to an instance of a
ProviderResource. 'providers', when passed to a component resource,
allows users to override providers used to construct children of the
component resource.
3. 'protect', 'providers', and 'provider' are all now inherited from
a resource's parent if they aren't specified in the child.
This commit adds the requisite code for the above changes and, in
addition, adds a number of new tests that exercise them and related code
paths.
* Rebase against master
* Add 'Output.all' combinator for Python
Output.all is a useful combinator that we already have in Node that
allows the composition of a list of outputs into an output of a list.
This is very useful when authoring components and its lack of presence
in Python was an oversight.
This commit adds 'Output.all' and 'Output.from_input', adding tests and
documentation for each.
* start unwrap
* Add functionality and test for nested inputs
* Fix, formalize and add tests for property rewrites
The Python SDK provides two hooks for resources to override how their
properties are communicated to and from the engine. The code that
performs this transformation is subtle and, before this commit, subtly
incorrect.
This commit adds a test that verifies that the SDK correctly transforms
properties recursively according to the two transformation hooks, while
also fixing a smattering of test issues encountered when adding the new
test.
* CR feedback
* Implement RegisterResourceOutputs for Python 3
RegisterResourceOutputs allows Python 3 programs to export stack outputs
and export outputs off of component resources (which, under the hood,
are the same thing).
Adds a new integration test for stack outputs for Python programs, as
well as add a langhost test for register resource outputs.
Fixespulumi/pulumi#2163
* CR: Rename stack_output -> export
Fix integration tests that hardcoded paths to stack_outputs
* Fix one more reference to stack_outputs
future_input tests that it's possible to use coroutines as inputs to
Pulumi resources. resource_thens tests that it's possible to use outputs
to chain resource inputs and outputs together and that the SDK reports
correct dependencies to the engine.
This PR also fixes two bugs exposed by the new tests: first, coroutines
needed to be scheduled before awaiting (otherwise we'd deadlock) and
Nones in maps needed to be ignored when serializing and deserializing.
* Implement RPC for Python 3
* Try not setting PYTHONPATH
* Remove PYTHONPATH line
* Implement Invoke for Python 3
* Implement register resource
* progress
* Rewrite the whole thing
* Fix a few bugs
* All tests pass
* Fix an abnormal shutdown bug
* CR feedback
* Provide a hook for resources to rename properties
As dictionaries and other classes come from the engine, the
translate_property hook can be used to intercept them and rename
properties if desired.
* Fix variable names and comments
* Disable Python integration tests for now
This commit introduces a 'next' package which we can use as a staging
ground for incrementally adopting new Python 3 code. The next package is
initially populated with the non-runtime portions of the Python SDK,
which is enough to pass all tests when running on Python 3. Future
commits will reach further into the runtime.
* Test the Python language host end-to-end
This commit introduces an end-to-end language host testing framework for
the Python SDK, similar to what already exists for the Node SDK. The
real language host is used to run Pulumi programs written in Python
while mocking out the resource monitor.
* Add new tests
* Print out better diagnostics when the langhost fails to launch
* Use the in-tree executor for testing
* CR: Place tests and code being tested in the same directory for ease of understanding, add a README
* Turns out I misunderstood the semantics of resource registration - fix two tests so that they pass now and fix a few bugs in the test harness
* Fix a few issues with the Python language host
1. Fix an issue where the UNKNOWN sentinel was leaking into user
programs
2. Fix an issue where Protobuf types were leaking into user programs
In fixing this issues I also added a framework for writing tests against
the Python SDK.
* License headers, and adopt a more idiomatic testing pattern
* Additional idiomatic Python
* That's what I get for trying to be fancy (Travis CI python version is very old and does not respect this form)
* CR feedback: use more comprehensions, typo fix
* Break a circular dependency between resource, runtime.resource, and runtime.rpc
* Don't check in .vscode
* CR: sort inputs, rename global variable, add a test for CustomResource serialization
* Remove accidental code duplication