Documentation
¶
Overview ¶
Package runner provides optional tooling for running multiple projections and scaling them safely. This package is designed to be explicit, deterministic, and CLI-friendly without imposing framework behavior or automatic scheduling.
Index ¶
Constants ¶
This section is empty.
Variables ¶
var ( // ErrNoProjections indicates that no projections were provided to run. ErrNoProjections = errors.New("no projections provided") // ErrInvalidPartitionConfig indicates invalid partition configuration. ErrInvalidPartitionConfig = errors.New("invalid partition configuration") )
Functions ¶
This section is empty.
Types ¶
type ProjectionRunner ¶
type ProjectionRunner struct {
Projection projection.Projection
Processor projection.ProcessorRunner
}
ProjectionRunner pairs a projection with its processor. The processor is adapter-specific (postgres.Processor, mysql.Processor, etc.) and knows how to manage transactions and checkpoints for that storage type.
type Runner ¶
type Runner struct{}
Runner orchestrates multiple projections concurrently. It is storage-agnostic and works with any processor implementation.
Example with PostgreSQL:
store := postgres.NewStore(postgres.DefaultStoreConfig()) processor1 := postgres.NewProcessor(db, store, &config1) processor2 := postgres.NewProcessor(db, store, &config2)
runner := runner.New()
err := runner.Run(ctx, []runner.ProjectionRunner{
{Projection: &MyProjection{}, Processor: processor1},
{Projection: &MyOtherProjection{}, Processor: processor2},
})
func (*Runner) Run ¶
func (r *Runner) Run(ctx context.Context, runners []ProjectionRunner) error
Run runs multiple projections concurrently until the context is canceled. Each projection runs in its own goroutine with its processor. Returns when the context is canceled or when any projection returns an error.
If a projection returns an error, all other projections are canceled and the error is returned. This ensures fail-fast behavior.
This method is safe to call from CLIs and does not assume single-process ownership. Coordination happens via the processor's checkpoint management.
Source Files