A tag is the only thing that is a version. Between two of them the tree is a
build nobody released, and until now it went on claiming to *be* the release
it came after — which is how 0.1.4 came to name both the build before the
per-node digest and the build after, with the mismatch messages unable to tell
them apart.
`+dev` is a local version: it says "0.1.4, plus changes", it does not have to
guess what the next release will be numbered, and PyPI refuses one, so a dev
build cannot be published by accident. `scripts/release.sh` in the workspace
root writes it after each tag; this is that commit, by hand, for the release
that predates it.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01A9Hdrmf2cwNABCnE5x9UJa
Comparing the per-node digest against an engine that does not record one is
comparing against nothing, and reporting every node as changed on every sync
for ever — which is what a client newer than its engine did, since `NodeDef`
drops fields it has never heard of. A node is named now only when both sides
carry a digest, so a no-op sync is `unchanged` again and the signal one syncs
for is back.
That silence had also been the only sign of the mismatch, so sync now names
it: one line saying the engine stored no record of what a node's code reaches,
with both versions in it and what to run. Bumped to 0.1.6 — the digest changed
the stored document's shape, and a version that does not move makes two
different engines indistinguishable, which is the thing it was made
load-bearing for a day ago.
`— draft` was printed whenever there was simply nothing to publish, which
reads as work left unfinished. It is said only when a draft is genuinely
there, and `— published` when one was.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01A9Hdrmf2cwNABCnE5x9UJa
Three things the first export pass got wrong for a real study.
**Dotted paths.** A node returns a record, not a scalar — the numbers arrive
inside `final_metrics` — so `--metrics final_metrics.train_loss` yielded an
empty column and `--metrics final_metrics` yielded the whole record in one
cell. Both sides of the wide table now take dotted paths, and the defaults
reach the same depth: every number a result carries is a column named by its
path, and inputs are compared leaf by leaf, so two configurations differing in
one field give that field as the axis rather than two blobs that are merely
not equal. Lists stay whole — a curve belongs in the long table.
**`--list`.** Metric names are flow-qualified, so `--name train_loss` matched
nothing and said only that. `fluksio export metrics --list` prints the names
the selection carries, and an empty export made with `--name` points at it.
**A version to compare.** The CLI ships ahead of the engine and a stale one
answered a flat 404 with nothing anywhere in the API to tell how old it was.
The engine reports `version` on `/observability/summary`, `fluksio status`
prints it, and a 404 from export now names both versions — or says "older"
when the field itself predates the engine. Bumped to 0.1.5, which is what
makes the number worth reading.
Also formats `flow/metrics.py`, which had been committed unformatted and was
the last `ruff format --check` failure.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01A9Hdrmf2cwNABCnE5x9UJa
`fluksio export metrics` is the long table — a row per run, metric and step —
and `fluksio export runs` the wide one, a row per run with the inputs that
*vary* across the selection as columns beside its final numbers, status,
duration and the commit and digest of the code it ran. Both carry the run id
on every row, which is the join back to the run page and what makes an
exported file auditable. `Client.export_metrics`/`export_runs` answer the same
rows to a notebook.
The engine streams csv or jsonl from two routes declared above `/{run_id}`;
parquet is a client-side conversion behind the new `fluksio[parquet]` extra,
so nobody pays for pyarrow who does not want dtypes kept. The long export
reads each run through `_series`, so a cached node's curve comes with it, and
`--stride` thins each series rather than the concatenation of all of them.
Two things they needed on the way: `GET /runs` takes `?since=` and `?before=`,
so a long history pages by the last row's own timestamp instead of an offset
that shifts under it; and a read that reaches no engine now says so in half a
second rather than seven, because `runs`, `flavors`, `export` and an unwatched
`status` pass `retries=0`. Everything that submits keeps them.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01A9Hdrmf2cwNABCnE5x9UJa
A driver script died of one slow answer: httpx.ReadTimeout out of
RunHandle.refresh() with a 30 s read timeout and no retry anywhere, which
cost a sweep 78 of its 84 runs.
- Split the timeout (5 s connect, 120 s read): a wrong URL fails at once,
and a busy engine gets longer than the slowest thing it does on purpose
(a 60 s compile, a 15 s rebuild wait).
- Retry idempotent calls three times on a transport error or 502/503/504.
503 is the engine's own "ask again" — it is what RebuildBusy answers.
- Submit carries a key the engine stores with the run, so a retry after a
timeout returns that run instead of starting a second. A sweep keys every
entry, so a half-created one recreates only what is missing.
- wait() and --follow tolerate five failed polls in a row; a 404 still stops
at once, because that is an answer rather than a gap.
- CLI says "engine not answering" and names the run still on the engine,
instead of printing a traceback.
- runs: clamp the params column to 80 characters; events() takes the
flow/since/until the endpoint already had; RunHandle.failures answers
"what killed this run" from the run's own node rows.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The engine was I/O-bound on its own state backend. `RedisState.lock()` is one
key — `pipeline:_lock` — for the whole process, taken five times a message at
two round trips each, and every cascade and every node read queued behind it.
Inside it, reading a node's inputs was three round trips per input (an EXISTS
for `in`, then EXISTS and GET for the value), writing was two updates that a
single transaction already gives, and the version counters went one INCR at a
time.
Replaced with the atomic command that was always available: `get_present` is
one MGET and tells a missing key from one holding null, so the lock it used to
be read under bought nothing; value and timestamp land in one `update`, which
is a MULTI/EXEC; `increment_multi` pipelines the counters. `values()` — what
every websocket snapshot calls — is two reads whatever the message count
instead of two per message.
Beside that: every webhook did its blocking XADD on the asyncio event loop
(MQTT already used `to_thread`); the per-execution `NodeOutcome` was built and
validated even with no run watching; `_minute` built a tz-aware datetime per
event on the loop thread to key a dict, and now keys on an int; `move_due`
promoted delayed items one round trip each, every second; `FLOW_MAX_CASCADES`
makes the in-flight ceiling a setting rather than a constant.
`orjson` replaces stdlib json where a message pays for it — state, the
journal, the engine side of the worker pipe. `fluksio-worker` stays
dependency-free, and the run-cache digest stays on stdlib so no stored key is
invalidated. A non-finite number now stores as `null` rather than the bare
`NaN` that was never JSON.
Measured with `scripts/bench_engine.py` against a real Redis, 200 messages:
a five-node chain went from 43.9 to 103.1 msg/s with p50 latency 2110ms →
782ms and p95 3913ms → 1439ms; one source into twenty consumers went from 5.4
to 33.7 msg/s. In memory, twenty consumers went from 187 to 448 msg/s.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01BpfSinyCBfjuieikyfMPbf
Two halves of the same gap: the CLI could start work but not show you any.
`fluksio status` draws the home screen's top half in a terminal — health and
what is wrong with it, every flow with its state and node count, and the
recent runs and failures under them. `--watch` keeps it there. Rich does the
drawing; it was already installed under fastapi's own CLI, and is named now
because a command depends on it.
`fluksio run` with no parameters at a terminal asks for them, one line per
declared input with its declared value in brackets — so Enter through the lot
is what running the defaults looks like, and an artifact input takes the
`@run:` spelling the engine now resolves. A scripted run is untouched: passing
any parameter, or piping the command, skips the questions, as does --defaults.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_019V5bsYGNxcgPs4xXmTPx69
The gates have never gone green on the new runners. Three separate reasons:
- backend/Dockerfile shipped Python 3.10 while the code imports typing.Self
and datetime.UTC, so the container exited on import and the suite could not
even load its conftest. The image moves to 3.13 and the packages declare
>=3.12, which is the floor the tests actually pass on; ruff's target follows
and rewrites timezone.utc and asyncio.TimeoutError accordingly. Relocking
drops the 3.10 branch, which bumps FastAPI and so regenerates the SDK.
- frontend/README.md had no trailing newline and two dashboard widgets used
arbitrary text-[…] sizes. Both are em-relative on purpose, so they move to
the inline style the neighbouring ramp already uses.
- Every commit left its own run queued: without a concurrency group a runner
that was offline for a while works through a backlog nobody reads. A stack
that fails to come up now prints its logs before the teardown removes it.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Replace the empty LICENSE placeholders with the verbatim GNU AGPL v3 text,
fix the invalid "AGPLv3" SPDX string in the package metadata, and name
Melvin Strobl as the copyright holder wherever the old footers said
"Fluksio ... all rights reserved".
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The release workflow was already right; what it would have uploaded was not.
`fluksio` had no readme, so its PyPI page would have been blank — the app
repo's own README is a contributor's map of `frontend/` and `docker/`, which
is the wrong front page for `pip install fluksio`. It now has one of its own,
aimed at somebody who landed on the project page. Both distributions gain
authors, urls, keywords and classifiers; `twine check` passes clean on all
four artifacts where it warned on two before.
The workflow publishes `fluksio-worker` first, because `fluksio` depends on it
and the other order leaves a few seconds — the whole of a first release — in
which the dependency cannot be resolved. `--check-url` makes a re-run skip
what is already uploaded rather than failing on it, which matters because a
version on PyPI can never be replaced.
Licence metadata is deliberately still absent: LICENSE is MIT in somebody
else's name, inherited from the template this was scaffolded from, and whose
it should be is not a decision to make in a commit. NOTEPAD carries it.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012ue1tkFWB1bcGy3aWhCKpU
A data scientist keeps their code where it is and decorates it: `@node`
declares a function's ports beside the function, `Flow(name, nodes=[...])`
says which of them make a flow, and `use(fn, wire=..., **settings)` rebinds
one for a single flow. `fluksio sync` uploads the document plus a generated
import shim per node, so the store still holds a complete, runnable,
git-versioned definition while the code it imports stays theirs.
`fluksio login|run|runs` and `flow.submit().wait()` are the client half, over
the run endpoints that already existed. Runs record the user repository's
commit beside the store's, so "what code produced this number" is answerable
on the side that now holds the code.
- `fluksio/sdk/`: ports, decorators, the flow builder and its checks, the shim
generator, an HTTP client and sync. Standard library only at import, so
`from fluksio import node` in a training script pulls in no engine.
- `FlowDef.origin` marks a flow code-defined; `Run.origin_commit` carries the
repository's commit; `POST /modules/refresh` retires the workers without an
install, which every sync calls — a worker holds the imported package in
memory, so an edit to it is invisible until the process goes.
- The canvas shows a generated body read-only and names the repository to edit
instead; a body edited there stops the next sync rather than being discarded.
- The worker's reporter carries inert `Port`, `node`, `use` and `Flow`, since
the shim imports a module whose first line declares them.
- `examples/myresearch` is the worked example, `make sync-example` uploads it.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012ue1tkFWB1bcGy3aWhCKpU
`pip install fluksio && fluksio serve` on a machine with no Docker, no
database and no configuration — which is the case this is for: a node on
a cluster where ports cannot be opened. It makes its data directory, its
key and an admin account, prints the password once, and serves. Pairing
is `fluksio enroll <code> --portal …`, doing what the Settings screen
does through the same function, before the engine starts and without one
running — a machine nobody can route to has no browser pointed at it
either. The portal serves the dashboard, so nothing is served here.
Two things had to give way. `fastapi[standard]` pulls a cloud CLI that
wants sentry-sdk 2.x while we pinned below it — no pip resolution
existed, so the pin is lifted, which the comment beside it had been
waiting for and which also lets the Python cap go. And `uv` is now a
dependency rather than something to find on PATH: the Modules screen is
how a data scientist installs torch, and it was quietly falling back to
the engine's own interpreter.
The CLI imports nothing from the engine before it has set DATA_DIR — the
settings are built on the first import of core.config, and reaching it
early put the database in the working directory. There is a test for
that now, because the failure is silent.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
One process owns this database — the image has run a single uvicorn
worker for that reason since the four-engines bug — so a file beside the
flows is the honest shape for it, and it is what lets `fluksio serve`
need no infrastructure at all. Live values, node execution and the work
queue never came here anyway; what does is a rollup a minute at a time,
a row per cascade and the run history, and WAL keeps the readers going
while that one writer works.
DATA_DIR is now the one setting that moves everything an installation
keeps; the rest derive from it and the images still spell theirs out.
The schema is prepared in-process at startup, so the prestart service is
gone, and the ten Postgres-only revisions collapse into one portable
baseline.
Three things only worked because psycopg was casting for us: a token's
subject arriving as a string where the column is a UUID, `greatest`, and
`date_bin`. The timestamps needed a column type of their own — SQLite
stores no offset, and a naive datetime read back either raises against an
aware `now` or serialises as local time.
Postgres stays in the stack only for Umami, behind the analytics profile.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
A cluster or GPU host installs `pip install fluksio-worker` and gets the
agent and the runner, not psycopg, numpy and the MCP SDK. The engine
depends on it as a workspace member, so the file it launches node code
with is the same file a remote worker runs — which is what keeps a node
unable to tell the difference. Copying the two files by hand still works.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
A wheel whose top-level module is `app` collides with anything else in a
user's venv, so the package that is about to be published takes the name
it is published under. Only the Python package moves; the repo, the
Docker WORKDIR and the compose project keep theirs.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The engine runs where the automations are and the GPU is somewhere else,
usually behind a different network — so the worker connects out and the engine
answers over the socket it was given. Nothing has to expose Redis, and the
same connection works through the tunnel the hosted access will use.
What travels is the protocol the local pool already speaks, so a node cannot
tell which kind of worker it is on. A node declares device: gpu and
device_policy, the label is resolved per call (a worker attaching later needs
no rebuild), and a run whose labels nothing carries waits in the queue saying
what it waits for rather than failing — submit from the couch, the GPU box
picks it up when it is switched on.
Two things had to move with it. Compiling now happens on the machine that will
run the node: a node importing torch is correct on the GPU box and a missing
module on the engine, so checking it here failed nodes that were fine. And the
artifact endpoint accepts a worker's own credential, because storing a
checkpoint is exactly what that credential is for — and only that.
Verified against the real split: the training ran on this host (its checkpoint
names the machine and a numpy the engine does not have), streamed 40 metric
points back mid-run, and the evaluate node read the checkpoint on the engine.
Cancel kills the remote training; pulling the worker fails the run in six
seconds instead of waiting out its ten-minute timeout.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AD8SfVhzXBG2nAfFcVh3iD
The worker script is handed to the interpreter by path, so app/flow was
sys.path[0] for every node: `import queue` got the engine's. It now drops
its own directory before anything else imports, and runs with the
deployment's credentials scrubbed out of its environment.
Also: reload builds off the event loop, the pool wakes what is blocked on
it when it stops, a refused metrics flush is kept for the next one rather
than dropped, and the cascade events are paired through failures.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017MeiWk3Yq12n2pTvnQWYvt
mqtt, http, influx and delay carried the prototype's annotations and were
excluded wholesale. Annotating them leaves nothing to exclude: what the two
untyped dependencies need is an implicit-reexport allowance for
influxdb_client and the import-site ignore croniter already gets elsewhere.
InfluxDbNode.inject now takes the durable flag its supertype passes, and the
cron loop reads its stop event once instead of dereferencing a field that
stop_cron() may have cleared.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01H7LwYgJfpkbLCTeiAf8U4A
nodes.py had grown to 2k lines holding every integration behind a single
blanket mypy exemption. It is now a package split by the outside world
each node talks to, so the exemption shrinks to the four integration
modules; base and mlp are type-checked, which turned up a dozen missing
annotations.
The senders opened a fresh connection — and, in the MQTT case, a fresh
thread pool and event loop — for every single message. HTTP senders now
share one pooled client, and a publisher holds one broker connection for
its lifetime, fed from a bounded queue that drops the oldest value when
the broker cannot keep up.
An HTTP sender also no longer trips over a JSON reply that is not an
object: outputs are keyed by port, so a bare scalar is a valid reply with
nothing to publish.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011LF61rxW1FG5YCD2J9YqjY
The engine now speaks MCP at /mcp, with a built-in OAuth 2.1 authorization
server in front of it: an agent registers itself, sends a human to the browser
to approve it, and exchanges the resulting code for a token. PKCE is required,
codes are single-use and stored only as hashes, the browser is redirected to
the URI that was registered rather than the one asked for, and refresh tokens
rotate so that replaying a spent one revokes the whole line.
Twenty tools cover reading, building, publishing and running flows, and each
one calls the same REST endpoint the dashboard calls, in-process, carrying the
caller's own token. That keeps one description of what a flow is and what may
be done to it — validation, the draft/publish split, the version check — and
means an agent can do nothing a person could not do in the browser.
Agent tokens are RS256 with a keypair of their own rather than the secret that
signs browser sessions, so deleting the key withdraws every agent without
logging anyone out, and deps.decode_token grew the branch that trusting a
second issuer will need when the hosted login arrives.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Flows can now be taken off the engine and put back. Stopped state lives in a
runtime.json beside the flow, not in the flow document: the canvas autosaves
that document, so a stopped flow would otherwise start itself again on the
next edit. A stopped flow gets no subscriptions, schedules or webhooks, its
nodes are skipped by the scheduler, and running it answers 409. Pausing holds
a flow's nodes while its values keep arriving, so the canvas still shows what
is coming in.
Node code is user code and print is how it says things, so stdout is teed
through a contextvar sink active only during a node execution — one event per
execution, capped, so a chatty node cannot outrun the stream. A node that
fails sends its traceback the same way, trimmed to the author's own frames.
The dock gains a logs panel and a pause control; the dashboard replaces its
placeholder with what is running, stopped or failing; the edge inspector can
send the last message again.
Single-stepping is deferred and noted: the scheduler keeps no progress between
calls, so a step button would re-run the same node rather than advance.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The engine fell back to MemoryState because there was no Redis to talk
to, so every value was lost on restart. Adds a redis service with
append-only persistence, reachable only from the backend, and points
REDIS_HOST at it.
state.py and pipeline.py leave the strict-mypy exclusion. redis-py types
its calls as sync/async unions; the casts narrow that at the points of
use rather than switching to redis.asyncio, which would make a
deliberately synchronous backend asynchronous. nodes.py stays excluded.
The .env.example CORS entry belongs with the CI work that follows.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01KkmeRiyeYmVZqJVwuyHq9o
Makes the flow engine reachable from the API, which is what M3 needs before
any of it can reach the browser.
- app/flow is a package now; the prototype's watch-dir scripts and the
matplotlib/networkx visualiser are gone with their dependencies.
- Messages carry a serializable dtype instead of a live Python type, and a
port name, so the graph can speak qualified names while node functions keep
local arguments. Redis state is JSON, not pickle.
- Message names are namespaced per flow ("heating.temp"); a bare name resolves
to its own flow, a dotted one crosses flows.
- Several nodes may provide the same message: producers are a list, so fan-in
is a real edge instead of a silently dropped one.
- Flows are stored as flow.json plus node sources in a git repository, one
commit per save, with identical saves skipped so autosave stays quiet.
- Node failures are isolated and reported per node; validate() returns cycles
and unconnected inputs instead of raising deep in a run.
- Credentials live in an encrypted store and are referenced as {"$secret": …}.
- Engine events reach websocket clients through a bus, so values, node status
and execution show up live.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016WzrvW7rjQbynnhF6pxh6i
- exclude app/flow from ruff and mypy: it is not importable as a package and
nothing in the API reaches it, so 405 of the 412 findings were about code
scheduled for restructuring (see NOTEPAD.md)
- fix the 7 real findings outside it: col(...) around created_at for the
.desc() ordering, and a type: ignore that is no longer needed
- compose.local.yml resets the host ports compose.dev.yml publishes; the
integrated stack is served entirely through Traefik, and a port already
taken on the host used to fail the whole stack
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The submodule collapse was only half applied: .gitmodules was deleted but
backend/ and frontend/ were still recorded as gitlinks, so none of their
files were tracked. Replace the gitlinks with the real trees.
Also untrack .env (it carried placeholder secrets) in favour of a tracked
.env.example, drop the committed __pycache__, and narrow the blanket *.png
ignore that would have swallowed design assets.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
stroblmeandClaude Opus 5