A scaffold example application built on top of MicroPhoenix.

The app can run on host BEAM (mix run) or AtomVM (./bin/atomvm-run). Both use port 8080 — do not run them at the same time.

mix deps.get
mix run --no-halt

The server will start on http://localhost:8080/

Req.get uses raw socket access through ExTCP.
If :eperm occurs, grant cap_net_raw to beam.smp:

ERL_ROOT=$(erl -noshell -eval 'io:format("~s",[code:root_dir()]), halt().')
sudo setcap cap_net_raw+ep "$ERL_ROOT"/erts-*/bin/beam.smp
getcap "$ERL_ROOT"/erts-*/bin/beam.smp

Expected output includes:

.../beam.smp cap_net_raw=ep

If you see _build/... permission denied, fix ownership:

sudo chown -R "$(id -un)":"$(id -gn)" _build

Before starting the server, you can create 5 sample Items.Item records with:

mix run -e "Enum.each(MicroScaffoldExample.Seed.names(), fn name -> MicroScaffoldExample.Items.create_item(name) end)"

Requires PostgreSQL and psql. See config/config.exs for connection settings.

Build and run on AtomVM:

export ATOMVM_INSTALL_PREFIX=/path/to/AtomVM/build
export PATH="$ATOMVM_INSTALL_PREFIX/src:$PATH"

mkdir -p avm_deps   # must be empty
mix deps.get
mix atomvm.packbeam
./bin/atomvm-run

bin/atomvm-run loads stdlib AVMs (atomvmlib, estdlib, exavmlib) together with the app.

On AtomVM, /index2.html lists 5 seeded items (compile-time, same names as host).

Proxies http://localhost:8000/index3.html via Req.get. On AtomVM, Req uses gen_tcp (not ExTCP raw sockets). Start an upstream HTTP server on port 8000 before testing:

# example upstream (separate terminal)
python3 -m http.server 8000 --directory /path/with/index3.html

If upstream is down, /index3.html returns HTTP 502.

Host BEAM still uses ExTCP and needs cap_net_raw (see above).