4269 Commits (efa85192ef35c542dfe4b0cacaa157fc5df9d460)

Author	SHA1	Message	Date
James Bardin	efa85192ef	PrepareComputedBlocks test	4 weeks ago
James Bardin	0cab7be239	fix internal validate	4 weeks ago
James Bardin	5410f9bd6a	objchange.PrepareComputedBlocks ... PrepareComputedBlocks nulls out any empty containers in a cty.Value corresponding to computed blocks, to indicate to the provider that the value is safe to insert a computed value into. TODO: do we really need this, or should the framework take care of it like it currently does already? - Doing this makes the protocol consistent that only null values can be replaced with computed values, but it's also more work that is currently redundant.	4 weeks ago
James Bardin	f133e0380a	change BlockByPath to NestedBlockByPath ... The function is only ever concerned with nested blocks, and we can make use of the nesting type information for computed block transformations.	4 weeks ago
James Bardin	9ae47096ae	update objchange validations ... add computed block handling to AssertPlanValid and AssertObjectCompatible.	4 weeks ago
James Bardin	0f2503e4ea	detect non-legacy schemas ... We already had a heuristic for flagging non-legacy schemas in the blocktoattr fixup code, but now we need to extend that to computed blocks. Adapt these checks to the block schema type itself since we want to use the check outside of the old blocktoattr package.	4 weeks ago
James Bardin	46d537b71b	add computed flag to schema blocks	4 weeks ago
James Bardin	73c225dff4	Merge pull request #37986 from hashicorp/jbardin/planned-private-for-random ... Resource pre-planned private data	4 weeks ago
James Bardin	32ff98ebee	Merge pull request #38298 from hashicorp/jbardin/terraform-data-extensions ... terraform_data write-only and sensitive extensions	4 weeks ago
James Bardin	7f653e1548	full apply test for private plan data	4 weeks ago
Roniece	dc4e328e62	Restore apply behavior	4 weeks ago
James Bardin	f28c6ee74f	make protobuf	1 month ago
James Bardin	1f4bf797ff	add ability to forward PlannedPrivate data ... Terraform providers sometimes want to be able to plan a computed field which may not be deterministically generated. For example, a random GUID could be generated during plan, making the overall Terraform plan more precise, but there is currently no way to carry that planned value forward to apply. The problem is that there are always two entirely independent plans for every resource change. Terraform compares these two plans for consistency, so a provider may not return a different computed value, but there is no way for the provider to see what the prior planned value was. This means that a planned value which is not reproducible from the given inputs is not plan-able at all. This PR allows Terraform to take the private data from the first plan, which is usually discarded, and send it along in the second plan request. The provider can then use that data to recreate any previously planned values, either because they are stored directly, or contain enough seed data to recreate the same result.	1 month ago
James Bardin	b477351cf0	update schemas where they are referenced in various tests ... Update a stacks test to ensure we create up-to-date values for the terraform_data resource. We also remove the encoded test values for easier reading. Update the copy of the schema necessary for the rpcapi test	1 month ago
James Bardin	641b820baa	terraform_data sensitive and write-only tools ... Add a new object to the terraform_data resource to handle sensitive and write-only attributes. The new `store` object has input and output values, which work much like the top-level input and output. The difference is that the `store.input` is truly "write only" and can also accept ephemeral values. Additional arguments for the `store` object are: - `version`, allows the user to determine exactly when the input value will be replaced in `store.output`. - `sensitive`, conditionally directs the stored value between `output` or `sensitive_output`. - `replace` signals that any change to either output should trigger replacement of the entire resource.	1 month ago
Sarah French	2dbb7d9c05	PSS: Remove automatic creation of the default workspace's state file during `init` (#38281) ... * refactor: Stop Terraform creating the default workspace during init when using PSS This is part of reconciling how in the past backends always reported that the default backend existed, even when it didn't. We want state stores to report reality only, so we need to let Terraform handle the discrepancy. Prior to this commit we handled it by making reality match the old lie that the default workspace always exists. After this commit we're just embracing Terraform working with truthful information.	1 month ago
Lakshay Oza	fd6b53b4f1	Fix grammar in`textdecodebase64` error message (#38289) ... * Fix grammar in base64 error message * Add changelog entry for base64 error message fix * Remove changelog entry	1 month ago
mickael-hc	332a94d9a1	update gpg with latest version from website (#38271) ... * update gpg with latest version updated expiry to 2030-03-01, key can be found at https://www.hashicorp.com/.well-known/pgp-key.txt * update another reference * include both old and new pubkeys ensures the certification of the old key is still present	1 month ago
Daniel Banck	977a0fb117	Potentially fetch backend variables for most commands ... * modules * providers * providers lock * providers mirror * validate * state mv * state pull * state push * state replace provider * state rm * taint * get	1 month ago
Daniel Banck	5bf90e356f	Allow commands to fetch backend variables ... The new `resolveConstVariables` helper checks if const variables are present in the configuration, but don't have a value yet. In that case we try to fetch them via a backend. This will only work for backends that implement the ConstVariableSupplier interface. (Which the `cloud` backend does) The precedence is the same as for existing commands: a CLI supplied value will override a value from a workspace	1 month ago
Daniel Banck	36b5207d2b	Still require const variables when AllowUnsetVariables is set ... We never want to stub const variables, so we will always try to get values for them. The cloud backend now always fetches variable values so const vars can be satisfied.	1 month ago
Daniel Banck	1d517c3ae8	Add check for missing const variables	1 month ago
Daniel Banck	97534ebd3b	Improve ParseVariableValues signature with a helper ... Instead of passing a "magic" boolean as third parameter, we now have two functions `ParseConstVariableValues` and `ParseVariableValues`.	1 month ago
Daniel Banck	0fa9e5b4da	command: Add vars to get command (and refactor it)	1 month ago
Daniel Banck	cdbd4f17f2	command: Add vars to taint command	1 month ago
Daniel Banck	a836cd610d	command: Add vars to state rm command	1 month ago
Daniel Banck	2115032765	command: Add vars to state replace command	1 month ago
Daniel Banck	36b26535b9	command: Add vars to state push command	1 month ago
Daniel Banck	aac8834600	command: Add vars to state pull command	1 month ago
Daniel Banck	b2a5ce8af1	command: Add vars to state mv command	1 month ago
Daniel Banck	c28c6c6d64	command: Add vars to providers schema command	1 month ago
Daniel Banck	ed2bdf6825	command: Add vars to providers mirror command	1 month ago
Daniel Banck	6c6be5bcfa	command: Add vars to pvoiders lock command	1 month ago
Daniel Banck	1cc8345926	command: Add vars to providers command	1 month ago
Daniel Banck	c9e1b3a47b	command: Add vars to modules command	1 month ago
Daniel Banck	a1c74c7d4e	command: Add vars to graph command	1 month ago
Daniel Banck	4b132a4879	command: Add vars to validate command	1 month ago
Daniel Schmidt	2cddb70898	use go-cty's WrangleMarksDeep instead of UnmarkDeepWithPaths ... This should improve the performance since we don't need to remark the values again and only handle the marks we want to deal with.	1 month ago
Daniel Banck	2767639191	Test cases for new typed module evaluations	1 month ago
James Bardin	9495ff1aa4	more precise output evaluation with types ... Allow for more precise types to be used when evaluating unknown output values. If we know all the types of outputs, and there are no dynamic types involved, we can use lists and maps for the modules, which allows module values to better match between validation and plan/apply. We also fix an old problem of modules evaluating to lists and maps during validation but not plan/apply, and causing expressions to evaluate differently in different phases. We use the existence of a type declaration to opt-in to this very subtly different behavior. It's not expected that any config would notice the difference since we are only removing possible type mismatches.	1 month ago
Daniel Banck	45ccfbf87a	Test output type override merge	1 month ago
Daniel Banck	8c88364e1c	Add evaluation scope and expr to output diagnostic	1 month ago
Daniel Banck	8b3329e910	Check type contraints for output values	1 month ago
Daniel Banck	941dfcc82c	Add `type` attribute to `output` blocks	1 month ago
Sarah French	d7baa140f8	fix: Add missing or correct values to error diagnostics raise when initialising a state store from backend state file data (#38275)	1 month ago
Daniel Banck	a47b89170f	Fix ParseVariableValues for different command flows ... We want to treat `const` variables as part of the configuration, we always require a value for them when loading configuration. Since most command load configuration via `Meta.loadConfig` or `Meta.loadConfigWithTests`, these flows call `ParseVariableValues` with `constOnly` true. Backend operations like `plan` and `apply` require values for all required variables (as before), so here we set `constOnly` to false, to keep the existing logic.	1 month ago
James Bardin	dea9f7ec38	Merge pull request #38270 from hashicorp/jbardin/update-cty ... update go-cty@v1.18	1 month ago
Roniece	88f8567b77	Update copyright headers	1 month ago
Roniece	87b37486b9	Restore applying.go	1 month ago
Roniece	d9045c8b97	Mirror existing functions and parse with stackaddrs.ParseActionInvocationInstanceStr	1 month ago