Even if you made a custom formatter for gorm, it would blithely println
anything returned, including nils, resulting in lots of new lines when
we were ignoring output. This also creates a custom logger that doesn't
do anything when newlines are found.
* database: add wt_public_id domain type
* static-hosts: add migrations
* Rename migrations
* migrations: add description column and fix foreign keys
* Add protobuf for static host
* Add HostCatalog and HostSet
* Add Options
* Add foreign key constraint
* Add package doc
* Add copied scanner code
* Add NewHostCatalog with tests
* Override default table name for static host catalog
* WIP: save static host catalog to database
* WIP: Print db url for debugging
* Add iam test helper function
* Make gorm default to null for names and descriptions
* Add foreign key constraint from host catalog to scope
* Names are not needed for the test org and project
* Avoid an ordering dependency in down migrations
* Use wt_timestamp
* Add check constraint on host address column
* Add Host
* Add HostSet
* Add HostSetMember
* Test a HostSetMember can be written to the database
* Move the getters and setters for table name to a new file
The TableName methods are needed by gorm and the SetTableName methods
are for oplog. Moving them to a new file so related functionality is in
the same file.
* Add repository
* Add NewRepository
* Close connection before cleanup of database
* Add comments to protobuf file and regenerate
* Define repository methods for HostCatalog
* Assign PublicId when resource is inserted into the database
* Wrap a nil parameter error
* Add oplog tickets for static host types
* WIP: first cut at CreateCatalog
* CreateHostCatalog: more tests
* Detect unique constraint violation
* Use const for PublicId prefixes
* Use PostgreSQL error name instead of error code
* Add TODO comments
* Fix comments
* Use PublicId not PublicID
* Refactor: move error variables from host/static to db
The error variables defined in host/static are generic and can be reused
by other packages.
* Create a new repository for each test
* Validate CreateCatalog returns a new HostCatalog
* Do not modify the HostCatalog passed in
* Implement UpdateCatalog
* Minor cleanup to make folds work
* Implement LookupCatalog
* Refactor: extract allocCatalog function
* Add comment
* Refactor: extract newCatalogMetadata function
* Make error strings more consistent
* Move database specific error check to internal/db package
* Change signature of DeleteCatalog and update docs
* Implement DeleteCatalog
* Make error strings more consistent
* Add examples
* Update docs
* Add package level doc and example for repository
* No need for using reflect.DeepEqual
* Remove redundant comment
* Clarify ErrMultipleRecords results in the transaction being rolled back
* Remove redundant error checks
* Remove unnecessary check
* Return early instead of using nested 'else' statements
* Update and Delete must return a row count
* catch err on defers (#82)
Co-authored-by: Jim Lambert <jimlambrt@Jims-MBP-3.home>
* Remove 'if assert' pattern
* Use assert package in test helper function
* Add test for updating a non existent catalog
* Use a single assert when a specific error is expected
* Use assert package when it clarifies the check
* Catch errors in cleanup functions
* Update comments in protobuf file
* Update doc for UpdateCatalog to reflect the new behavior
* Add common error values for field mask errors
* Require a field mask when calling update methods
* Add TODO for removing direct call to gorm
* Export Public ID prefixes
* Add db timestamp guards
* Remove unused option
Co-authored-by: Jim <jlambert@hashicorp.com>
Co-authored-by: Jim Lambert <jimlambrt@Jims-MBP-3.home>
* carve out scope into a separate branch for a PR
* refactor getOpts to be private
* narrowed to just the scope proto
* wrap in a transaction
* remove user from scopes PR
* make scope type immutable
* refactored with more referential integrity (thanks mike)
* refactor based on pair programming with Mike
* better VetForWrite that supports db.Options but that requires GetOpts and Options to be exported.
* stop printing out all the db logs
* refactor withScope to be private
* refactor the VetForWrite a bit based on Todd's PR comment
* fix typo
Co-authored-by: Michael Gaffney <mgaffney@users.noreply.github.com>
* fix ctx propagation and Scope returns
* fix ctx propagation and use a const for std retry count
* change Std to Crud
* fix ctx propagation
* make sure update with field masks doesn't update a field not in the mask and force a read after update.
* add description
* remove comment about future
* clean error strings
* clean up enum string rep
* stdMeta
* fix return value to be just nil
* fix returns and errors
* update test to match new error string
* change len of domain to 24
* provide common function for new public ids
* use common db.NewPublicId() and provide the correct prefix
* move list out of CrudActions and provide a CrudlActions shortcut as well
* some comments about the purpose of VetForWrite()
* move to random 10
* Adding project API interface.
* Creation of Project handler and Tests
* Move to using the iam objects.
* Setting up testify framework for mocking and assertions.
* Fix iam options typo references.
* Adding tests for CreateProject and UpdateProject.
This also adds some additional argument validation.
* Merging in PR 40
* Renamed fakeRepo to mockRepo
* Changing around TODOs to capture current plans with the code.
* Remove the repo interface and use a real DB for tests.
* Fixed a few tests.
* Adding checks for properly formatted identifiers.
* Updating code to account for recent merges. Relaxing some restrictions on what an id can look like.
* Added small comment explaining why RegisterGrpcGatewayer interface exists.
* Move project service up a call level.
Long term I think it would make sense to define these services even higher up the call stack, potentially at the initialization of the controller server, near where the DB is initialized.
* Making these fields align with those in the resources under hosts/...
* Hard code the service initialization in the handler registration code.
* Regenerate project related code proto code.
* Removing type field for project.
* Adds basic project handling in CLI/SDK (#47)
* Adds SDK for creating and reading projects
* Reorganizes a bit
* Adds CLI for creating and reading projects
* Fixes some some generation bugs
* Adds redirect support for default org ID
* Fix protobuf import references being moved.
* Update client test to look for "orgs" and "projects" instead of "org" and "project"
* Update expected segments test
* Fix setAddr handling when default organization is in use
* Set client in target after create or read
* Adding functionality and tests for update masks.
* Additional tests for update. Disable update non existant project test.
* Make not providing an update mask be an invalid argument.
* Make empty update mask paths and no mutable fields in the path an InvalidArgument error.
* Fix get Project test.
* Adding back in the Update a non existant project test case.
* Adding seperate path structured tests.
* Attach an issue id to field mask validation TODO.
* Generate Project Update SDK code (#58)
* Created the UpdateFunction method.
* Regenerate protobuf derived code since pull #50 updated dependencies.
* Special case IDs to not be passed for updates.
* Updated the tools file since imports reordered it.
* Add spaces between each tools import/comment grouping.
Co-authored-by: Jim Lambert <jimlambrt@Jims-MBP-3.home>
Co-authored-by: Jim <jlambert@hashicorp.com>
Co-authored-by: Michael Gaffney <mgaffney@users.noreply.github.com>
Co-authored-by: Jeff Mitchell <jeffrey.mitchell@gmail.com>
This can be used in tests to get access to a controller that behaves
just like a dev controller but doesn't require the normal Command bits
(UI, etc).
This can be enhanced over time as needed; for now it's pretty simple but
it's good enough to be used for API tests. A simple project create/read
API test is added as well, which serves to show the programmatic model
(as well as test those functions at the API level).
* Add wt_timestamp type to database
* Swap the order of domains and oplog migrations
All migrations should use the new 'wt_timestamp' domain type for
timestamps.
* Remove dead code
* Add down migration for wt_timestamp
Jim
Michael Gaffney
Todd Knight
Jeff Mitchell
