# Upgrading from 0.3

This page is for you if you copied modules out of `examples/` in 0.3: the
worker, the worker kernel, the reaper or a language adapter. In 0.4 those
modules ship with the application under `wasm_` names. Nothing you copied
breaks: no new module shares a name with anything in the 0.3 `examples/`, so
your copies keep compiling and running. Moving to the installed modules is
optional, and you can do it one module at a time.

## Pick your case

| you copied | do this |
| --- | --- |
| `wasm_worker.erl`, unchanged | switch the calls to `wasm_instance_worker`, below, and delete your copy |
| `wasm_worker.erl`, changed | keep it, or move to `wasm_instance_worker` if its options now cover your change |
| the kernel: `script_worker`, `worker_reaper`, `script_v1`, `worker_error`, an adapter | switch with the table below when you are ready |

## Switch the simple worker

`wasm_instance_worker` takes the 0.3 example's calls exactly: `start_link/1,2`,
`call/3,4`, `stop/1` and `info/1`, with the same options and the same return
values. So switching is a rename of the calls:

```sh
perl -pi -e 's/\bwasm_worker:/wasm_instance_worker:/g' $(grep -rlw wasm_worker src)
```

Then delete `src/wasm_worker.erl`. If your copy was renamed, as
`getting-started.md` used to suggest, put your module's name in place of
`wasm_worker` in that command.

## Switch the kernel

| 0.3, copied | 0.4, installed |
| --- | --- |
| `worker_reaper:start_link(#{scratch => Dir})` in your supervision tree | remove it: the application runs the reaper. To keep recovery across node restarts, set `{wasm, [{scratch_roots, #{scratch => Dir}}]}` |
| `worker_reaper:start_link(Roots, Settings)` | `{wasm, [{scratch_roots, Roots}, {reaper_options, Settings}]}` |
| `script_worker:start_link(Adapter, #{root => scratch, ...})` | `wasm_script_worker:start_link(Adapter, #{...})`; `root` defaults to `scratch` |
| `js_worker:run(W, Src, Ctx)`, `python_worker:run(W, Src, Ctx)` | `wasm_script_worker:run(W, Src, Ctx)` |
| `qjs_adapter`, `qjs_reactor_adapter` | `wasm_javascript_command`, `wasm_javascript` |
| `py_adapter`, `py_reactor_adapter` | `wasm_python_command`, `wasm_python` |
| `lua_reactor_adapter` | `wasm_lua` |
| `-behaviour(script_worker)` in your own adapter | `-behaviour(wasm_worker_adapter)`; the callbacks are unchanged |
| `worker_error:runtime(E)` and the other constructors | `wasm_worker_error:runtime(E)` |
| `worker_reaper:stats/0`, `requests/0` | `wasm_script_worker:cleanup_stats/0`, `cleanup_requests/0` |

Limits and option keys are otherwise unchanged. The profile is still called
`script_v1` and the boot scripts are still in `priv/script_v1/`, so a guest
built for 0.3 needs nothing.

## Where scratch files go now

Without `scratch_roots`, the reaper keeps request files in a directory of this
node's own under the user cache directory, and removes it at a clean shutdown.
If the node crashes, that directory is not reclaimed on the next start, since
the next start uses a new one. Set `scratch_roots` in production: a configured
root is where a restarted node finds and cleans what a crashed one left. A
configured root must belong to one node at a time.

## Delete what you replaced

Remove the copies you switched away from, run your tests, and check nothing
still names them:

```sh
grep -rlwE 'script_worker|worker_reaper|script_v1|qjs_adapter|py_adapter' src
```