| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| | |
|
| |
|
|
|
|
| |
- bug doesn't commit identities anymore, only make sure they are commit
- cache use an IdentityResolver to load bugs with identities from the cache (deps injection)
- IdentityCache now are identity.Interface
|
| |
|
|
|
|
|
|
|
| |
- allow the creation of arbitrary Lamport clocks, freeing the way to new entities and removing Bug specific (upper layer) code.
- generalize the memory-only and persisted Lamport clocks behind a common interface
- rework the tests to provide reusable testing code for a Repo, a Clock, a Config, opening a path to add a new Repo implementation more easily
- test previously untested components with those new tests
Note: one problem found during this endeavor is that `identity.Version` also need to store one time + Lamport time for each other Entity (Bug, config, PR ...). This could possibly done without breaking change but it would be much easier to wait for https://github.com/MichaelMure/git-bug-migration to happen.
|
| | |
|
| | |
|
| |
|
|
|
|
| |
Turns out it was a mistake, who is suprised ?
It leaks the abstraction and it's generally much cleaner to let the cache layer make both calls
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Some git operations require the user to have an identity configured and
will exit with failure if none is set (or if git can't guess it). As a
direct consequence of this, the test suite may fail depending on the
user local configuration.
The error itself is justified as regular users *should* configure their
identity themselves. However, when building in chrooted environments
it's unlikely the git identity will be set making the test suite fail
unnecessarily.
To prevent such unnecessary failures, let's make a dummy identity for
repos created and used by the test suite.
|
| |
|
|
| |
metadata later
|
| | |
|
| |
|
