Via a new accumulative `--tpt-proto` arg you can select which
`tpt_protos: list[str]`-fixture protocol keys will be delivered to
opting in tests!
B)
Also includes,
- CLI quote handling/stripping.
- default of 'tcp'.
- only support one selection per session at the moment (until we figure
out how we want to support multiples, either simultaneously or
sequentially).
- draft a (masked) dynamic-`metafunc` parametrization in the
`pytest_generate_tests()` hook.
- first proven and working use in the `test_advanced_faults`-suite (and
thus its underlying
`examples/advanced_faults/ipc_failure_during_stream.py` script)!
|_ actually needed this to prove that the suite only has 2 failures on
'uds' seemingly due to low-level `trio` error semantics translation
differences to do with with calling `socket.close()`..
On a very nearly related topic,
- draft an (also commented out) `set_script_runtime_args()` fixture idea
for a std way of `partial`-ling in runtime args to `examples/`
scripts-as-modules defining a `main()` which would proxy to
`tractor.open_nursery()`.
Since in hindsight the real analog of a net-proto's "bindspace"
(normally its routing layer's addresses-port-set) is more akin to the
"location in the file-system" for a UDS socket file (aka the file's
parent directory) determines whether or not the "port" (aka it's
file-name) collides with any other.
So the `._filedir: Path` is like the allocated "address" and,
the `._filename: Path|str` is basically the "port",
at least in my mind.. Bp
Thinking about fs dirs like a "host address" means you can get
essentially the same benefits/behaviour of say an (ip)
addresses-port-space but using the (current process-namespace's)
filesys-tree. Note that for UDS sockets in particular the
network-namespace is what would normally isolate so called "abstract
sockets" (i.e. UDS sockets that do NOT use file-paths by setting `struct
sockaddr_un.sun_path = 'abstract', see `man unix`); using directories is
even easier and definitely more explicit/readable/immediately-obvious as
a human-user.
As such this reworks all the necessary `UDSAddress` meths,
- `.unwrap()` now returns a `tuple(str(._filedir, str(._filename))`,
- `wrap_address()` now matches UDS on a 2nd tuple `str()` element,
- `.get_root()` no longer passes `maybe_pid`.
AND adjusts `MsgpackUDSStream` to,
- use the new `unwrap_sockpath()` on the `socket.get[sock/peer]name()`
output before passing directly as `UDSAddress.__init__(filedir, filename)`
instead of via `.from_addr()`.
- also pass `maybe_pid`s to init since no longer included in the
unwrapped-type form.
Such that the "global-ish" setting (actor-local) is managed with the
others per actor-process and type it as a `Literal['tcp', 'uds']` of the
currently support protocol keys.
Here obvi `_tpt` is some kinda shorthand for "transport" and `_proto` is
for "protocol" Bp
Change imports and refs in all dependent modules.
Oh right, and disable UDS in `wrap_address()` for the moment while
i figure out how to avoid the unwrapped type collision..
There was a very strange legacy test
`test_spawning.test_local_arbiter_subactor_global_state` which was
causing unforseen hangs/errors on the UDS tpt and looking deeper this
test was already doing root-actor things that should never have been
valid XD
So rework that test to properly demonstrate something of value
(i guess..) and add a new suite which start more rigorously auditing our
`open_root_actor()` permitted usage.
For the old test,
- since the main point of this test seemed to be the ability to invoke
the same function in both the parent and child actor (using the very
legacy `ActorNursery.run_in_actor()`.. due to be deprecated) rename it
to `test_run_in_actor_same_func_in_child`,
- don't re-enter `.open_root_actor()` since that's invalid usage (tested
in new suite see below),
- adjust some `spawn()` arg/var naming and ensure we only return in the
child.
For the new suite add tests for,
- ensuring the implicit `open_root_actor()` call under `open_nursery()`.
- double open of `open_root_actor()` from within the same process tree
both from a root and sub.
Intro some new `_exceptions` used in the new suite,
- a top level `RuntimeFailure` for generically expressing faults not of
our own doing that prevent successful operation; this is what we now
(changed in this commit) raise on attempts to open a 2nd root.
- mk `ActorFailure` derive from the former; it's already used from
`._spawn` when subprocs fail to boot.
- aggregate the `MsgStream.aclose()` "reader tasks" stats content into a
common `message: str` before emit.
- tweak an `_rpc.process_messages()` emit per new `Channel.__repr__()`.
Throughout `_context` and `_spawn` where it causes no big disruption.
Still lots to work out for things like how to pass `--uid
<tuple-as-str>` to spawned subactors and whether we want a diff name for
the minimum `tuple` required to distinguish a subactor pre-process-ID
allocation by the OS.
Finally.. i've been meaning todo this for ages since the
actor-id-swap-as-handshake is better layered as part of the IPC msg-ing
machinery and then let's us encapsulate the connection-time-assignment
of a remote peer's `Aid` as a new `Channel.aid: Aid`. For now we
continue to offer the `.uid: tuple[str, str]` attr (by delegating to the
`.uid` field) since there's still a few things relying on it in the
runtime and ctx layers
Nice bonuses from this,
- it's very easy to get the peer's `Aid.pid: int` from anywhere in an
IPC ctx by just reading it from the chan.
- we aren't saving more then the wire struct-msg received.
Also add deprecation warnings around usage to get us moving on porting
the rest of consuming runtime code to the new attr!
For the case where there's clearly no socket file created/bound
obviously the `trio.socket.connect()` call will raise
`FileNotFoundError`, so just translate this to
a builtin-`ConnectionError` at the transport layer so we can report the
guilty `UDSAddress`.
The more I think about it, it seems @guille's orig approach of
unwrapping UDS socket-file addresses to strings (or `Path`) is making
the most sense. I had originally thought that pairing it with the
listening side's pid would add clarity (and it definitely does for
introspection/debug/logging) but since we don't end up passing that pid
to the eventual `.connect()` call on the client side, it doesn't make
much sense to wrap it for the wire just to discard.. Further, the
`tuple[str, int]` makes `wrap_address()` break for TCP since it will
always match on uds first.
So, on that note this patch refines a few things in prep for going back
to that original `UnwrappedAddress` as `str` type though longer run
i think the more "builtin approach" would be to add `msgspec` codec
hooks for these types to avoid all the `.wrap()`/`.unwrap()` calls
throughout the runtime.
Down-low deats,
- add `wrap_address()` doc string, detailed (todo) comments and handle
the `[None, None]` case that can come directly from
`._state._runtime_vars['_root_mailbox']`.
- buncha adjustments to `UDSAddress`,
- add a `filedir`, chng `filepath` -> `filename` and mk `maybe_pid` optional.
- the intent `filedir` is act as the equivalent of the host part in a network proto's
socket address and when it's null use the `.def_bindspace = get_rt_dir()`.
- always ensure the `filedir / filename` is an absolute path and
expose it as a new `.sockpath: Path` property.
- mk `.is_valid` actually verify the `.sockpath` is in the valid
`.bindspace: namely just checking it's in the expected dir.
- add pedantic `match:`ing to `.from_addr()` such that we error on
unexpected `type(addr)` inputs and otherwise parse any `sockpath:
Path` inputs using a new `unwrap_sockpath()` which simply splits an
abs file path to dir, file-name parts.
- `.unwrap()` now just `str`-ifies the `.sockpath: Path`
- adjust `.open/close_listener()` to use `.sockpath`.
Such that we can run (opting-in) tests on both TCP and UDS backends and
ensure the `reg_addr` fixture and various timeouts are adjusted
accordingly.
Impl deats,
- add a new `tpc_proto` CLI option and fixture to allow choosing which
"transport protocol" will be used in the test suites (either globally
or contextually).
- rm `_reg_addr` instead opting for a `_rando_port` which will only be
used for `reg_addr`s which are net-tpt-protos.
- rejig `reg_addr` fixture to set a ideally session-unique `testrun_reg_addr`
based on the `tpt_proto` setting making appropriate calls to `._addr`
APIs as needed.
- refine `daemon` fixture a bit with typing, `tpt_proto` timings, and
stderr capture.
- in `test_discovery` do a ton of type-annots, add `debug_mode` fixture
opt ins, augment `spawn_and_check_registry()` with `psutil.Process`
passing for introspection (when things go wrong..).
Call it `maybe_block_bp()` can wrap the `open_root_actor()` body with
it. Main reason is to guarantee we can bp inside actor runtime bootup as
needed when debugging internals! Prolly should factor this to another
module tho?
ALSO, ensure we RTE on recurrent entries to `open_root_actor()` from
within an existing tree! There was actually `test_spawning` test somehow
getting away with this!? Should never be possible or allowed!
And map `.__repr__/__str__` to it and add various new fields to fill it
out,
- drop `self.uid` as var and instead add `Actor._aid: Aid` and proxy to
it for the various `.name/.uid/.pid` properties as well as a new
`.aid` field.
|_ the `Aid.pid` addition is also included.
Other improvements,
- flip to a sync call to `Address.close_listener()`.
- track the `async_main()` parent task as `Actor._task`.
- add exception logging around failure to bind due to already-in-use
when calling `add.open_listener()` in `._stream_forever()`; sometimes
the error might be overridden by something else during the
runtime-failure unwind..
Namely reducing the duplication of class-fields and `TypeVar`s used
for parametrizing the `Address` protocol type,
- drop all of the `TypeVar` types and just stick with all concrete addrs
types inheriting from `Address` only.
- rename `Address.name_key` -> `.proto_key`.
- rename `Address.address_type` -> `.unwrapped_type`
- rename `.namespace` -> `.bindspace` to better reflect that this "part"
of the address represents the possible "space for binding endpoints".
|_ also linux already uses "namespace" to mean the `netns` and i'd
prefer to stick with their semantics for that.
- add `TCPAddress/UDSAddress.def_bindspace` values.
- drop commented `.open_stream()` method; never used.
- simplify `UnwrappedAdress` to just a `tuple` of union types.
- add logging to `USDAddress.open_listener()` for now.
- adjust `tractor.ipc/_uds/tcp` transport to use new addr field names.
Much like we already do in the `._iter_packets()` async-generator which
delivers to `.recv()` and `async for`, handle the `''[Errno 32] Broken
pipe'` case that can show up with unix-domain-socket usage.
Seems like the cause is due to how fast the socket can be torn down
during a registry addr channel ping where,
- the sending side can break the connection faster then the pong side
can prep its handshake msg,
- the pong side tries to send it's handshake pkt via
`.SocketStream.send_all()` after the breakage and then raises
`trio.BrokenResourceError`.
Such that whenev the `self._ctx.chan._exc is trans_err` we suppress.
I.e. when the `Channel._exc: Exception|None` error **is the same as**
set by the `._rpc.process_messages()` loop (that is, set to the
underlying transport layer error), we suppress the lowlevel tb,
otherwise we deliver the full tb since likely something at the lowlevel
that we aren't detecting changed/signalled/is-relevant!
Much like how `Context` has been implemented, try to give tons of high
level details on all the lower level encapsulated primitives, namely the
`.msgstream/.transport` and any useful runtime state.
B)
Impl deats,
- adjust `.from_addr()` to only call `._addr.wrap_address()` when we
detect `addr` is unwrapped.
- add another `log.runtime()` using the new `.__repr__()` in
`Channel.from_addr()`.
- change to `UnwrappedAddress` as in prior commits.
Previously whenever an `ActorNursery.start_actor()` call did not receive
a `bind_addrs` arg we would allocate the default `(localhost, 0)` pairs
in the parent, for UDS this obviously won't work nor is it ideal bc it's
nicer to have the actor to be a socket server (who calls
`Address.open_listener()`) define the socket-file-name containing their
unique ID info such as pid, actor-uuid etc.
As such this moves "random" generation of server addresses to the
child-side of a subactor's spawn-sequence when it's sin-`bind_addrs`;
i.e. we do the allocation of the `Address.get_random()` addrs inside
`._runtime.async_main()` instead of `Portal.start_actor()` and **only
when** `accept_addrs`/`bind_addrs` was **not provided by the spawning
parent**.
Further this patch get's way more rigorous about the `SpawnSpec`
processing in the child inside `Actor._from_parent()` such that we
handle any invalid msgs **very loudly and pedantically!**
Impl deats,
- do the "random addr generation" in an explicit `for` loop (instead of
prior comprehension) to allow for more detailed typing of the layered
calls to the new `._addr` mod.
- use a `match:/case:` for process any invalid `SpawnSpec` payload case
where we can instead receive a `MsgTypeError` from the `chan.recv()`
call in `Actor._from_parent()` to raise it immediately instead of
triggering downstream type-errors XD
|_ as per the big `#TODO` we prolly want to take from other callers
of `Channel.recv()` (like in the `._rpc.process_messages()` loop).
|_ always raise `InternalError` on non-match/fall-through case!
|_ add a note about not being able to use `breakpoint()` in this
section due to causality of `SpawnSpec._runtime_vars` not having
been processed yet..
|_ always return a third element from `._from_rent()` eventually to be
the `preferred_transports: list[str]` from the spawning rent.
- use new `._addr.mk_uuid()` and pass to new `Actor.__init__(uuid: str)`
for all actor creation (including in all the mods tweaked here).
- Move to new type-alias-name `UnwrappedAddress` throughout.
Such that any UDS socket pair is represented (and with the recent
updates to) a `USDAddress` via a similar pair-`tuple[str, int]` as TCP
sockets, a pair of the `.filepath: Path` & the peer proc's `.pid: int`
which we read from the underlying `socket.socket` using
`.set/getsockopt()` calls
Impl deats,
- using the Linux specific APIs, we add a `get_peer_info()` which reads
the `(pid, uid, gid)` using the `SOL_SOCKET` and `SOL_PEECRED` opts to
`sock.getsockopt()`.
|_ this presumes the client has been correspondingly configured to
deliver the creds via a `sock.setsockopt(SOL_SOCKET, SO_PASSCRED,
1)` call - this required us to override `trio.open_unix_socket()`.
- override `trio.open_unix_socket()` as per the above bullet to ensure
connecting peers always transmit "credentials" options info to the
listener.
- update `.get_stream_addrs()` to always call `get_peer_info()` and
extract the peer's pid for the `raddr` and use `os.getpid()` for
`laddr` (obvi).
|_ as part of the new impl also `log.info()` the creds-info deats and
socket-file path.
|_ handle the oddity where it depends which of `.getpeername()` or
`.getsockname()` will return the file-path; i think it's to do with
who is client vs. server?
Related refinements,
- set `.layer_key: int = 4` for the "transport layer" ;)
- tweak some typing and multi-line unpacking in `.ipc/_tcp`.
A few things that can fundamentally change,
- UDS addresses now always encapsulate the local and remote pid such
that it denotes each side's process much like a TCP *port*.
|_ `.__init__()` takes a new `maybe_pid: int`.
|_ this required changes to the `.ipc._uds` backend which will come in
an subsequent commit!
|_ `UDSAddress.address_type` becomes a `tuple[str, int]` just like the
TCP case.
|_ adjust `wrap_address()` to match.
- use a new `_state.get_rt_dir() -> Path` as the default location for
UDS socket file: now under `XDG_RUNTIME_DIR'/tractor/` subdir by
default.
- re-implement `USDAddress.get_random()` to use both the local
`Actor.uid` (if available) and at least the pid for its socket file
name.
Removals,
- drop the loop generated `_default_addrs`, simplify to just
`_default_lo_addrs` for per-transport default registry addresses.
|_ change to `_address_types: dict[str, Type[Address]]` instead of
separate types `list`.
|_ adjust `is_wrapped_addr()` to just check `in _addr_types.values()`.
- comment out `Address.open_stream()` it's unused and i think the wrong
place for this API.
Renames,
- from `AddressTypes` -> `UnwrappedAddress`, since it's a simple type
union and all this type set is, is the simple python data-structures
we encode to for the wire.
|_ see note about possibly implementing the `.[un]wrap()` stuff as
`msgspec` codec `enc/dec_hook()`s instead!
Additions,
- add a `mk_uuid()` to be used throughout the runtime including for
generating the `Aid.uuid` part.
- tons of notes around follow up refinements!
Type hint all ctx managers in _ringbuf.py
Remove unnecesary send lock on ring chan sender
Handle EOF on ring chan receiver
Rename ringbuf tests to make it less redundant
Add EOF signaling mechanism
Support proper `receive_some` end of stream semantics
Add StapledStream non-ipc test
Create MsgpackRBStream similar to MsgpackTCPStream for buffered whole-msg reads
Add EventFD.read cancellation on EventFD.close mechanism using cancel scope
Add test for eventfd cancellation
Improve and add docstrings
Add trio.StrictFIFOLock on sender.send_all
Support max_bytes argument on receive_some, keep track of write_ptr on receiver
Add max_bytes receive test test_ringbuf_max_bytes
Add docstrings to all ringbuf tests
Remove EFD_NONBLOCK support, not necesary anymore since we can use abandon_on_cancel=True on trio.to_thread.run_sync
Close eventfd's after usage on open_ringbuf
Tyler Goodlet
Guillermo Rodriguez