This change adds the following:
- Space UID + Custom harness validation (accountId for top level space, harness identifier for child spaces)
- Repo UID + Custom harness validation (harness identifier)
- Store Unique casing of space / repo path and add Path.ValueUnique (with Unique index) to allow for application layer controlling the case sensitivity (case insensitive standalone vs partially case sensitive harness)
- Token UID (unique index over ownertype + ownerID + tokenUID)
- Add DisplayName for principals (replaces Name to avoid confustion)
- Store Unique casing of principal UID and add Principal.ValueUnique (with unique index) to allow for application layer, per principal type control of case sensitivity (required in embedded mode)
- Generate serviceAccount UID (+Email) Randomly (sa-{space|repo}-{ID}-{random}) - Allows to have a unique UID across all principals while reducing likelyhood of overlaps with users + avoid overlap across spaces / repos.
- Sync casing of space names (accountId orgId projectId) when creating spaces on the fly (to ensure case sensitivity of - harness code) or use the existing space to update casing.
- Update serviceaccount client to match updated NG Manager API
- in embedded mode create spaces for harness resources owning the service account
To simplify UI code we are going to expose both Harness API and Standalone API (restricted to harness embedded functionalities) when running in harness mode.
Furthermore, this PR adds a middleware that allows us to reuse standalone Handlers for Harness API for operations that don't require any request/response manipulation.
This commit contains the following:
- Improve and simplify error handling (remove unnecessary wrappers, make it feel like go)
- Add extra validation for path creation and resource moving (path has to be within same top space, no top space alias allowed)
- Add access logging for rest api and git api
This change is adding the concept of Paths.
A repository and space always have a Primary Path which always is represents the ancestry to the root space.
All access history / resource visibility / child listings / UI traversal / etc. is done via that path.
Additionally, repos and spaces can have Alias Paths, which as the name states are aliases. via the primary path.
They sole impact is that a space or repo can be reached via different paths from the UI / rest apis / git apis.
This fulfills two major purposes:
- Customers can rename or move projects and spaces without breaking any existing references from CI pipeliens / code bases / local repos / ...
- Customer can create shorter aliases for important repos when in harness embeded mode! (acc/org/proj/repo can be shortened to acc/repo, or acc/repo'
Apart from the path changes, this PR adds:
Improved User facing errors
Improved internal error handling and wrapping
update / rename operation for repo and space
path list / delete / create operation for repo and space
