Migrating from FLEX v1 to v2¶

FLEX v2 is a clean break: v1 stays on the main branch and keeps working; migrate scripts when convenient. This page maps every v1 concept to its v2 home.

Setup¶

v1	v2
clone repo + `pip install -e .` (one big package)	`irm flex.levylab.org/install.ps1 \\| iex`, then `flex ecosystem use levylab` for the lab stack
hardcoded DB / Nextcloud / n8n endpoints in code	one manifest in the repo's `ecosystems/` folder: `flex ecosystem show` after activating `levylab`
secrets in source	environment variables (`NEXTCLOUD_PASSWORD`, `ASANA_ACCESS_TOKEN`)

Imports and classes¶

v1	v2
`from flex.inst.base import Instrument` (ZMQ baked in)	`from flex_protocols import ZMQInstrument` (also `VISAInstrument`, `TCPInstrument`, `SerialInstrument`)
`from flex.inst.levylab.Lockin import Lockin`	`from flex_drivers.levylab.lockin import Lockin`
`inst._send_command(cmd, params)['result']`	`inst.call(cmd, params)` — errors now raise instead of returning `None`
`_LABVIEW_CLASS_NAME` module constant	`lv_class` class attribute; `flex_drivers.levylab.lvclass_registry()` derives the CESession lookup from it
`flex.inst.levylab.insttypes.*` empty mixins	`flex_drivers.levylab.capabilities` protocols (real methods; `exp.get(Temperature)`)
`from flex.exp.experiment import Experiment`	`from flex import Experiment`
`from flex.exp.CESession import CESession`	`from flex import CESession`
`flex.db.FLEXDB` (raw SQL, hardcoded host)	metadata store from config: SQLite default, `[db] backend = "postgres"` for the lab DB
`flex.tdms.flexTDMS.write_tdms`	`[data] writer = "tdms"` — every measurement writes TDMS automatically
`flex.nextcloud`	`[storage] backend = "nextcloud"` — files upload on measurement finish
`flex.exp.dbexptoAsana.trigger_n8n_dbexptoAsana()` (n8n) / `flex.asana.Asana` (hardcoded-token stub)	`[comms] backend = "asana"` — `flex-asana` talks to the Asana API directly, token from `ASANA_ACCESS_TOKEN`
`flex.exp.users` hardcoded `Literal` list	any user string; lab-side validation can hook `experiment.start`
`flex.exp.script_to_db.CellLogger`	built into `Experiment` (`cell_log=True`), stored in `flex_cells`

Method names¶

Driver methods are now snake_case; the wire protocol is unchanged: setAO_Amplitude(...) → set_ao_amplitude(...) (still sends setAO_Amplitude).

Behavior changes worth knowing¶

JSON-RPC errors raise ZMQInstrumentError (v1 silently returned None).
A ZMQ timeout no longer wedges the connection — the socket is reset and the next call works; you get a TimeoutError you can catch.
Ctrl-C during a Scan is safe: the data file is finalized, the measurement is marked aborted, and on_abort cleanups run.
Experiment/measurement IDs gained a random suffix (20260709153053-bee7) so two starts in the same second cannot collide.
Metadata store failures never kill a measurement — they are logged and the experiment continues (opt into strictness with [exp] strict_metadata = true).
The five PPMS* classes are one PPMS (they only differed in address); pass the address, or define stations in the config.

Not carried over (deliberately)¶

The v1 flex.asana.Asana file contains a live, hardcoded Asana token — revoke it (its v2 replacement reads ASANA_ACCESS_TOKEN from the environment instead; see the row above).
flex.lv LabVIEW call helpers, exp/pund.py (experiment logic, not framework — rewrite as a Scan script), exp/experiment_bak.
Newport/Ophir/Sphere drivers that require .NET or COM — tracked in packages/flex-drivers/DEFERRED.md.

Postgres note¶

PostgresStore creates the v2 core schema as flex_-prefixed tables (flex_experiments, flex_measurements, flex_notes, flex_cells, flex_logs, flex_instruments) and does not touch the v1 exp / meas / cell_log tables. Migrating that historical data (or pointing v2 at the same tables) is a coordinated one-off step — plan it with the lab before switching production.