authoring

package
v6.3.1 Latest Latest
Warning

This package is not in the latest version of its module.

Go to latest
Published: Nov 2, 2026 License: Apache-2.0 Imports: 7 Imported by: 0

README

Public workload examples

These packages use only the supported authoring API: pkg/bench, pkg/gen, pkg/report, and the root application shell. They require Go 1.27 or newer. To use them outside this repository, copy query.go, accounts.go, and mirror.go into your own importable package, change the package name as needed, and require github.com/stroppy-io/stroppy/v6. No internal imports or built-in workload dependencies are needed. The external-module test does exactly this against a release-shaped module archive, without a replace directive.

Run the example catalog from the repository:

go run ./examples/authoring/main probe -o json
go run ./examples/authoring/main run example/query -d noop --iterations 10
go run ./examples/authoring/main run example/accounts -d noop --iterations 10
go run ./examples/authoring/main run example/mirror \
  -dprimary noop -dsecondary noop --iterations 10 --report-format json
  • Query declares standard run parameters and executes a parameterized query.
  • Accounts loads ordinary struct rows through gen.FromRows, runs managed transfers with whole-body retry, verifies owned generic reads, and contributes a final report from typed metrics and the actual measurement window.
  • Mirror uses named driver references, a step-selected default database, a second database facade, finite metric labels, and a directly encoded payload. The two queries are independent, not an atomic distributed transaction.

Accounts creates and drops stroppy_example_accounts. Use a disposable database where that table does not already exist. Noop exercises generation, insertion, and transaction bookkeeping, but does not verify database contents. PostgreSQL and MySQL runs additionally read rows and check that transfers preserve the total balance:

go run ./examples/authoring/main run example/accounts -d pg \
  -D url='postgres://user:password@localhost:5432/scratch' \
  --rows 100 --load-workers 2 --iterations 20 --report-format json

Prepared initializes one dedicated session and two prepared statements per worker, reuses them across transactions, and validates each counter read before commit. It requires PostgreSQL or MySQL, not noop. It creates and drops stroppy_example_prepared; use a disposable database without that table. Pool capacity must cover every worker.

go run ./examples/authoring/main run example/prepared -d pg \
  -D url='postgres://user:password@localhost:5432/scratch' \
  -D postgres.maxConns=8 --vus 8 --iterations 100

A single-test entrypoint can be smaller:

func main() { stroppy.Main(authoring.Accounts) }

For the installed catalog, export a Test value from your package and register it in init with bench.Register(Test). Then stroppy build ./workload links it into the managed runtime. Explicit bench.NewCatalog does not require global registration.

See the authoring guide for input precedence, filters, cancellation/drain semantics, and the v6 compatibility boundary.

Documentation

Overview

Package authoring contains small workloads using only the supported author API.

Index

Constants

This section is empty.

Variables

View Source
var (
	Accounts = bench.Test{Name: "example/accounts", Define: defineAccounts}
)

Accounts creates its own table, loads rows, then transfers balances atomically.

View Source
var Mirror = bench.Test{Name: "example/mirror", Define: defineMirror}

Mirror sends the same statement to independent primary and secondary databases. It makes no distributed-transaction or atomicity guarantee.

View Source
var (
	Prepared = bench.Test{Name: "example/prepared", Define: definePrepared}
)
View Source
var Query = bench.Test{Name: "example/query", Define: defineQuery}

Functions

This section is empty.

Types

This section is empty.

Directories

Path Synopsis

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL