tractor/tractor/ipc/_chan.py

821 lines
24 KiB
Python
Raw Normal View History

Re-license code base for distribution under AGPL This commit obviously denotes a re-license of all applicable parts of the code base. Acknowledgement of this change was completed in #274 by the majority of the current set of contributors. From here henceforth all changes will be AGPL licensed and distributed. This is purely an effort to maintain the same copy-left policy whilst closing the (perceived) SaaS loophole the GPL allows for. It is merely for this loophole: to avoid code hiding by any potential "network providers" who are attempting to use the project to make a profit without either compensating the authors or re-distributing their changes. I thought quite a bit about this change and can't see a reason not to close the SaaS loophole in our current license. We still are (hard) copy-left and I plan to keep the code base this way for a couple reasons: - The code base produces income/profit through parent projects and is demonstrably of high value. - I believe firms should not get free lunch for the sake of "contributions from their employees" or "usage as a service" which I have found to be a dubious argument at best. - If a firm who intends to profit from the code base wants to use it they can propose a secondary commercial license to purchase with the proceeds going to the project's authors under some form of well defined contract. - Many successful projects like Qt use this model; I see no reason it can't work in this case until such a time as the authors feel it should be loosened. There has been detailed discussion in #103 on licensing alternatives. The main point of this AGPL change is to protect the code base for the time being from exploitation while it grows and as we move into the next phase of development which will include extension into the multi-host distributed software space.
2021-12-13 18:08:32 +00:00
# tractor: structured concurrent "actors".
# Copyright 2018-eternity Tyler Goodlet.
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU Affero General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU Affero General Public License for more details.
# You should have received a copy of the GNU Affero General Public License
# along with this program. If not, see <https://www.gnu.org/licenses/>.
2018-05-30 16:36:23 +00:00
"""
Inter-process comms abstractions
2021-06-30 17:47:07 +00:00
2018-05-30 16:36:23 +00:00
"""
from __future__ import annotations
2022-07-12 15:22:30 +00:00
from collections.abc import (
AsyncGenerator,
AsyncIterator,
)
Proto `MsgCodec`, an interchange fmt modify API Fitting in line with the issues outstanding: - #36: (msg)spec-ing out our SCIPP (structured-con-inter-proc-prot). (https://github.com/goodboy/tractor/issues/36) - #196: adding strictly typed IPC msg dialog schemas, more or less better described as "dialog/transaction scoped message specs" using `msgspec`'s tagged unions and custom codecs. (https://github.com/goodboy/tractor/issues/196) - #365: using modern static type-annots to drive capability based messaging and RPC. (statically https://github.com/goodboy/tractor/issues/365) This is a first draft of a new API for dynamically overriding IPC msg codecs for a given interchange lib from any task in the runtime. Right now we obviously only support `msgspec` but ideally this API holds general enough to be used for other backends eventually (like `capnproto`, and apache arrow). Impl is in a new `tractor.msg._codec` with: - a new `MsgCodec` type for encapsing `msgspec.msgpack.Encoder/Decoder` pairs and configuring any custom enc/dec_hooks or typed decoding. - factory `mk_codec()` for creating new codecs ad-hoc from a task. - `contextvars` support for a new `trio.Task` scoped `_ctxvar_MsgCodec: ContextVar[MsgCodec]` named 'msgspec_codec'. - `apply_codec()` for temporarily modifying the above per task as needed around `.open_context()` / `.open_stream()` operation. A new test (suite) in `test_caps_msging.py`: - verify a parent and its child can enable the same custom codec (in this case to transmit `NamespacePath`s) with tons of pedantic ctx-vars checks. - ToDo: still need to implement #36 msg types in order to be able to get decodes working (as in `MsgStream.receive()` will deliver an already created `NamespacePath` obj) since currently all msgs come packed in `dict`-msg wrapper packets.. -> use the proto from PR #35 to get nested `msgspec.Raw` processing up and running Bo
2024-03-26 19:50:47 +00:00
from contextlib import (
asynccontextmanager as acm,
contextmanager as cm,
)
import platform
from pprint import pformat
import struct
Improved log msg formatting in core As part of solving some final edge cases todo with inter-peer remote cancellation (particularly a remote cancel from a separate actor tree-client hanging on the request side in `modden`..) I needed less dense, more line-delimited log msg formats when understanding ipc channel and context cancels from console logging; this adds a ton of that to: - `._invoke()` which now does, - better formatting of `Context`-task info as multi-line `'<field>: <value>\n'` messages, - use of `trio.Task` (from `.lowlevel.current_task()` for full rpc-func namespace-path info, - better "msg flow annotations" with `<=` for understanding `ContextCancelled` flow. - `Actor._stream_handler()` where in we break down IPC peers reporting better as multi-line `|_<Channel>` log msgs instead of all jammed on one line.. - `._ipc.Channel.send()` use `pformat()` for repr of packet. Also tweak some optional deps imports for debug mode: - add `maybe_import_gb()` for attempting to import `greenback`. - maybe enable `stackscope` tree pprinter on `SIGUSR1` if installed. Add a further stale-debugger-lock guard before removal: - read the `._debug.Lock.global_actor_in_debug: tuple` uid and possibly `maybe_wait_for_debugger()` when the child-user is known to have a live process in our tree. - only cancel `Lock._root_local_task_cs_in_debug: CancelScope` when the disconnected channel maps to the `Lock.global_actor_in_debug`, though not sure this is correct yet? Started adding missing type annots in sections that were modified.
2024-02-19 17:25:08 +00:00
import typing
from typing import (
2022-07-12 15:22:30 +00:00
Any,
Callable,
2022-07-12 15:22:30 +00:00
runtime_checkable,
Protocol,
Type,
TypeVar,
)
2018-05-30 16:36:23 +00:00
2024-04-02 17:41:52 +00:00
import msgspec
from tricycle import BufferedReceiveStream
2018-05-30 16:36:23 +00:00
import trio
from tractor.log import get_logger
2024-04-02 17:41:52 +00:00
from tractor._exceptions import (
MsgTypeError,
Refine `MsgTypeError` handling to relay-up-on-`.recv()` Such that `Channel.recv()` + `MsgpackTCPStream.recv()` originating msg-type-errors are not raised at the IPC transport layer but instead relayed up the runtime stack for eventual handling by user-app code via the `Context`/`MsgStream` layer APIs. This design choice leads to a substantial amount of flexibility and modularity, and avoids `MsgTypeError` handling policies from being coupled to a particular backend IPC transport layer: - receive-side msg-type errors, as can be raised and handled in the `.open_stream()` "nasty" phase of a ctx, whilst being packed at the `MsgCodec`/transport layer (keeping the underlying src decode error coupled to the specific transport + interchange lib) and then relayed upward to app code for custom handling like a normal Error` msg. - the policy options for handling such cases could be implemented as `@acm` wrappers around `.open_context()`/`.open_stream()` blocks (and their respective delivered primitives) OR just plain old async generators around `MsgStream.receive()` such that both built-in policy handling and custom user-app solutions can be swapped without touching any `tractor` internals or providing specialized "registry APIs". -> eg. the ignore and relay-invalid-msg-to-sender approach can be more easily implemented as embedded `try: except MsgTypeError:` blocks around `MsgStream.receive()` possibly applied as either of an injected wrapper type around a stream or an async gen that `async for`s from the stream. - any performance based AOT-lang extensions used to implement a policy for handling recv-side errors space can avoid knowledge of the lower level IPC `Channel` (and-downward) primitives. - `Context` consuming code can choose to let all msg-type-errs bubble and handle them manually (like any other remote `Error` shuttled exception). - we can keep (as before) send-side msg type checks can be raised locally and cause offending senders to error and adjust before the streaming phase of an IPC ctx. Impl (related) deats: - obvi make `MsgpackTCPStream.recv()` yield up any `MsgTypeError` constructed by `_mk_msg_type_err()` such that the exception will eventually be relayed up to `._rpc.process_messages()` and from their delivered to the corresponding ctx-task. - in support of ^, make `Channel.recv()` detect said mtes and use the new `pack_from_raise()` to inject the far end `Actor.uid` for the `Error.src_uid`. - keep raising the send side equivalent (when strict enabled) errors inline immediately with no upward `Error` packing or relay. - improve `_mk_msg_type_err()` cases handling with far more detailed `MsgTypeError` "message" contents pertaining to `msgspec` specific failure-fixing-tips and type-spec mismatch info: * use `.from_decode()` constructor in recv-side case to inject the non-spec decoded `msg_dict: dict` and use the new `MsgCodec.pld_spec_str: str` when clarifying the type discrepancy with the offending field. * on send-side, if we detect that an unsupported field type was described in the original `src_type_error`, AND there is no `msgpack.Encoder.enc_hook()` set, that the real issue is likely that the user needs to extend the codec to support the non-std/custom type with a hook and link to `msgspec` docs. * if one of a `src_type/validation_error` is provided, set that error as the `.__cause__` in the new mte.
2024-04-09 14:36:25 +00:00
pack_from_raise,
TransportClosed,
_mk_send_mte,
_mk_recv_mte,
2024-04-02 17:41:52 +00:00
)
Proto `MsgCodec`, an interchange fmt modify API Fitting in line with the issues outstanding: - #36: (msg)spec-ing out our SCIPP (structured-con-inter-proc-prot). (https://github.com/goodboy/tractor/issues/36) - #196: adding strictly typed IPC msg dialog schemas, more or less better described as "dialog/transaction scoped message specs" using `msgspec`'s tagged unions and custom codecs. (https://github.com/goodboy/tractor/issues/196) - #365: using modern static type-annots to drive capability based messaging and RPC. (statically https://github.com/goodboy/tractor/issues/365) This is a first draft of a new API for dynamically overriding IPC msg codecs for a given interchange lib from any task in the runtime. Right now we obviously only support `msgspec` but ideally this API holds general enough to be used for other backends eventually (like `capnproto`, and apache arrow). Impl is in a new `tractor.msg._codec` with: - a new `MsgCodec` type for encapsing `msgspec.msgpack.Encoder/Decoder` pairs and configuring any custom enc/dec_hooks or typed decoding. - factory `mk_codec()` for creating new codecs ad-hoc from a task. - `contextvars` support for a new `trio.Task` scoped `_ctxvar_MsgCodec: ContextVar[MsgCodec]` named 'msgspec_codec'. - `apply_codec()` for temporarily modifying the above per task as needed around `.open_context()` / `.open_stream()` operation. A new test (suite) in `test_caps_msging.py`: - verify a parent and its child can enable the same custom codec (in this case to transmit `NamespacePath`s) with tons of pedantic ctx-vars checks. - ToDo: still need to implement #36 msg types in order to be able to get decodes working (as in `MsgStream.receive()` will deliver an already created `NamespacePath` obj) since currently all msgs come packed in `dict`-msg wrapper packets.. -> use the proto from PR #35 to get nested `msgspec.Raw` processing up and running Bo
2024-03-26 19:50:47 +00:00
from tractor.msg import (
_ctxvar_MsgCodec,
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
# _codec, XXX see `self._codec` sanity/debug checks
Proto `MsgCodec`, an interchange fmt modify API Fitting in line with the issues outstanding: - #36: (msg)spec-ing out our SCIPP (structured-con-inter-proc-prot). (https://github.com/goodboy/tractor/issues/36) - #196: adding strictly typed IPC msg dialog schemas, more or less better described as "dialog/transaction scoped message specs" using `msgspec`'s tagged unions and custom codecs. (https://github.com/goodboy/tractor/issues/196) - #365: using modern static type-annots to drive capability based messaging and RPC. (statically https://github.com/goodboy/tractor/issues/365) This is a first draft of a new API for dynamically overriding IPC msg codecs for a given interchange lib from any task in the runtime. Right now we obviously only support `msgspec` but ideally this API holds general enough to be used for other backends eventually (like `capnproto`, and apache arrow). Impl is in a new `tractor.msg._codec` with: - a new `MsgCodec` type for encapsing `msgspec.msgpack.Encoder/Decoder` pairs and configuring any custom enc/dec_hooks or typed decoding. - factory `mk_codec()` for creating new codecs ad-hoc from a task. - `contextvars` support for a new `trio.Task` scoped `_ctxvar_MsgCodec: ContextVar[MsgCodec]` named 'msgspec_codec'. - `apply_codec()` for temporarily modifying the above per task as needed around `.open_context()` / `.open_stream()` operation. A new test (suite) in `test_caps_msging.py`: - verify a parent and its child can enable the same custom codec (in this case to transmit `NamespacePath`s) with tons of pedantic ctx-vars checks. - ToDo: still need to implement #36 msg types in order to be able to get decodes working (as in `MsgStream.receive()` will deliver an already created `NamespacePath` obj) since currently all msgs come packed in `dict`-msg wrapper packets.. -> use the proto from PR #35 to get nested `msgspec.Raw` processing up and running Bo
2024-03-26 19:50:47 +00:00
MsgCodec,
types as msgtypes,
pretty_struct,
Proto `MsgCodec`, an interchange fmt modify API Fitting in line with the issues outstanding: - #36: (msg)spec-ing out our SCIPP (structured-con-inter-proc-prot). (https://github.com/goodboy/tractor/issues/36) - #196: adding strictly typed IPC msg dialog schemas, more or less better described as "dialog/transaction scoped message specs" using `msgspec`'s tagged unions and custom codecs. (https://github.com/goodboy/tractor/issues/196) - #365: using modern static type-annots to drive capability based messaging and RPC. (statically https://github.com/goodboy/tractor/issues/365) This is a first draft of a new API for dynamically overriding IPC msg codecs for a given interchange lib from any task in the runtime. Right now we obviously only support `msgspec` but ideally this API holds general enough to be used for other backends eventually (like `capnproto`, and apache arrow). Impl is in a new `tractor.msg._codec` with: - a new `MsgCodec` type for encapsing `msgspec.msgpack.Encoder/Decoder` pairs and configuring any custom enc/dec_hooks or typed decoding. - factory `mk_codec()` for creating new codecs ad-hoc from a task. - `contextvars` support for a new `trio.Task` scoped `_ctxvar_MsgCodec: ContextVar[MsgCodec]` named 'msgspec_codec'. - `apply_codec()` for temporarily modifying the above per task as needed around `.open_context()` / `.open_stream()` operation. A new test (suite) in `test_caps_msging.py`: - verify a parent and its child can enable the same custom codec (in this case to transmit `NamespacePath`s) with tons of pedantic ctx-vars checks. - ToDo: still need to implement #36 msg types in order to be able to get decodes working (as in `MsgStream.receive()` will deliver an already created `NamespacePath` obj) since currently all msgs come packed in `dict`-msg wrapper packets.. -> use the proto from PR #35 to get nested `msgspec.Raw` processing up and running Bo
2024-03-26 19:50:47 +00:00
)
2018-05-30 16:36:23 +00:00
log = get_logger(__name__)
_is_windows = platform.system() == 'Windows'
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
def get_stream_addrs(
stream: trio.SocketStream
) -> tuple[
tuple[str, int], # local
tuple[str, int], # remote
]:
'''
Return the `trio` streaming transport prot's socket-addrs for
both the local and remote sides as a pair.
'''
# rn, should both be IP sockets
lsockname = stream.socket.getsockname()
rsockname = stream.socket.getpeername()
return (
tuple(lsockname[:2]),
tuple(rsockname[:2]),
)
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
# from tractor.msg.types import MsgType
# ?TODO? this should be our `Union[*msgtypes.__spec__]` alias now right..?
# => BLEH, except can't bc prots must inherit typevar or param-spec
# vars..
MsgType = TypeVar('MsgType')
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
# TODO: break up this mod into a subpkg so we can start adding new
# backends and move this type stuff into a dedicated file.. Bo
#
2022-07-12 15:22:30 +00:00
@runtime_checkable
class MsgTransport(Protocol[MsgType]):
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
#
# ^-TODO-^ consider using a generic def and indexing with our
# eventual msg definition/types?
# - https://docs.python.org/3/library/typing.html#typing.Protocol
stream: trio.SocketStream
2021-12-02 17:34:27 +00:00
drained: list[MsgType]
def __init__(self, stream: trio.SocketStream) -> None:
...
# XXX: should this instead be called `.sendall()`?
async def send(self, msg: MsgType) -> None:
...
async def recv(self) -> MsgType:
...
def __aiter__(self) -> MsgType:
...
def connected(self) -> bool:
...
2021-12-02 17:34:27 +00:00
# defining this sync otherwise it causes a mypy error because it
# can't figure out it's a generator i guess?..?
def drain(self) -> AsyncIterator[dict]:
...
@property
2022-07-12 15:22:30 +00:00
def laddr(self) -> tuple[str, int]:
...
@property
2022-07-12 15:22:30 +00:00
def raddr(self) -> tuple[str, int]:
...
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
# TODO: typing oddity.. not sure why we have to inherit here, but it
# seems to be an issue with `get_msg_transport()` returning
# a `Type[Protocol]`; probably should make a `mypy` issue?
2022-07-12 15:22:30 +00:00
class MsgpackTCPStream(MsgTransport):
2022-02-07 17:47:50 +00:00
'''
A ``trio.SocketStream`` delivering ``msgpack`` formatted data
using the ``msgspec`` codec lib.
'''
layer_key: int = 4
name_key: str = 'tcp'
# TODO: better naming for this?
# -[ ] check how libp2p does naming for such things?
codec_key: str = 'msgpack'
def __init__(
self,
stream: trio.SocketStream,
prefix_size: int = 4,
# XXX optionally provided codec pair for `msgspec`:
# https://jcristharif.com/msgspec/extending.html#mapping-to-from-native-types
#
# TODO: define this as a `Codec` struct which can be
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
# overriden dynamically by the application/runtime?
codec: tuple[
Callable[[Any], Any]|None, # coder
Callable[[type, Any], Any]|None, # decoder
]|None = None,
) -> None:
2018-05-30 16:36:23 +00:00
self.stream = stream
2019-12-10 05:55:03 +00:00
assert self.stream.socket
2019-12-10 05:55:03 +00:00
# should both be IP sockets
self._laddr, self._raddr = get_stream_addrs(stream)
2019-12-10 05:55:03 +00:00
# create read loop instance
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
self._aiter_pkts = self._iter_packets()
2018-12-15 07:20:19 +00:00
self._send_lock = trio.StrictFIFOLock()
2018-05-30 16:36:23 +00:00
# public i guess?
2021-12-02 17:34:27 +00:00
self.drained: list[dict] = []
self.recv_stream = BufferedReceiveStream(
transport_stream=stream
)
self.prefix_size = prefix_size
Proto `MsgCodec`, an interchange fmt modify API Fitting in line with the issues outstanding: - #36: (msg)spec-ing out our SCIPP (structured-con-inter-proc-prot). (https://github.com/goodboy/tractor/issues/36) - #196: adding strictly typed IPC msg dialog schemas, more or less better described as "dialog/transaction scoped message specs" using `msgspec`'s tagged unions and custom codecs. (https://github.com/goodboy/tractor/issues/196) - #365: using modern static type-annots to drive capability based messaging and RPC. (statically https://github.com/goodboy/tractor/issues/365) This is a first draft of a new API for dynamically overriding IPC msg codecs for a given interchange lib from any task in the runtime. Right now we obviously only support `msgspec` but ideally this API holds general enough to be used for other backends eventually (like `capnproto`, and apache arrow). Impl is in a new `tractor.msg._codec` with: - a new `MsgCodec` type for encapsing `msgspec.msgpack.Encoder/Decoder` pairs and configuring any custom enc/dec_hooks or typed decoding. - factory `mk_codec()` for creating new codecs ad-hoc from a task. - `contextvars` support for a new `trio.Task` scoped `_ctxvar_MsgCodec: ContextVar[MsgCodec]` named 'msgspec_codec'. - `apply_codec()` for temporarily modifying the above per task as needed around `.open_context()` / `.open_stream()` operation. A new test (suite) in `test_caps_msging.py`: - verify a parent and its child can enable the same custom codec (in this case to transmit `NamespacePath`s) with tons of pedantic ctx-vars checks. - ToDo: still need to implement #36 msg types in order to be able to get decodes working (as in `MsgStream.receive()` will deliver an already created `NamespacePath` obj) since currently all msgs come packed in `dict`-msg wrapper packets.. -> use the proto from PR #35 to get nested `msgspec.Raw` processing up and running Bo
2024-03-26 19:50:47 +00:00
# allow for custom IPC msg interchange format
# dynamic override Bo
2024-04-02 17:41:52 +00:00
self._task = trio.lowlevel.current_task()
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
# XXX for ctxvar debug only!
# self._codec: MsgCodec = (
# codec
# or
# _codec._ctxvar_MsgCodec.get()
# )
2021-12-02 17:34:27 +00:00
async def _iter_packets(self) -> AsyncGenerator[dict, None]:
'''
Yield `bytes`-blob decoded packets from the underlying TCP
stream using the current task's `MsgCodec`.
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
This is a streaming routine implemented as an async generator
func (which was the original design, but could be changed?)
and is allocated by a `.__call__()` inside `.__init__()` where
it is assigned to the `._aiter_pkts` attr.
'''
decodes_failed: int = 0
while True:
try:
2024-04-18 19:40:26 +00:00
header: bytes = await self.recv_stream.receive_exactly(4)
except (
ValueError,
ConnectionResetError,
# not sure entirely why we need this but without it we
# seem to be getting racy failures here on
# arbiter/registry name subs..
trio.BrokenResourceError,
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
) as trans_err:
loglevel = 'transport'
match trans_err:
# case (
# ConnectionResetError()
# ):
# loglevel = 'transport'
# peer actor (graceful??) TCP EOF but `tricycle`
# seems to raise a 0-bytes-read?
case ValueError() if (
'unclean EOF' in trans_err.args[0]
):
pass
# peer actor (task) prolly shutdown quickly due
# to cancellation
case trio.BrokenResourceError() if (
'Connection reset by peer' in trans_err.args[0]
):
pass
# unless the disconnect condition falls under "a
# normal operation breakage" we usualy console warn
# about it.
case _:
loglevel: str = 'warning'
raise TransportClosed(
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
message=(
f'IPC transport already closed by peer\n'
f'x)> {type(trans_err)}\n'
f' |_{self}\n'
),
loglevel=loglevel,
) from trans_err
# XXX definitely can happen if transport is closed
# manually by another `trio.lowlevel.Task` in the
# same actor; we use this in some simulated fault
# testing for ex, but generally should never happen
# under normal operation!
#
# NOTE: as such we always re-raise this error from the
# RPC msg loop!
except trio.ClosedResourceError as closure_err:
raise TransportClosed(
message=(
f'IPC transport already manually closed locally?\n'
f'x)> {type(closure_err)} \n'
f' |_{self}\n'
),
loglevel='error',
raise_on_report=(
closure_err.args[0] == 'another task closed this fd'
or
closure_err.args[0] in ['another task closed this fd']
),
) from closure_err
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
# graceful TCP EOF disconnect
if header == b'':
raise TransportClosed(
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
message=(
f'IPC transport already gracefully closed\n'
f')>\n'
f'|_{self}\n'
),
loglevel='transport',
# cause=??? # handy or no?
)
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
size: int
size, = struct.unpack("<I", header)
2021-09-05 20:29:16 +00:00
log.transport(f'received header {size}') # type: ignore
2024-04-18 19:40:26 +00:00
msg_bytes: bytes = await self.recv_stream.receive_exactly(size)
2021-09-05 20:29:16 +00:00
log.transport(f"received {msg_bytes}") # type: ignore
try:
Proto `MsgCodec`, an interchange fmt modify API Fitting in line with the issues outstanding: - #36: (msg)spec-ing out our SCIPP (structured-con-inter-proc-prot). (https://github.com/goodboy/tractor/issues/36) - #196: adding strictly typed IPC msg dialog schemas, more or less better described as "dialog/transaction scoped message specs" using `msgspec`'s tagged unions and custom codecs. (https://github.com/goodboy/tractor/issues/196) - #365: using modern static type-annots to drive capability based messaging and RPC. (statically https://github.com/goodboy/tractor/issues/365) This is a first draft of a new API for dynamically overriding IPC msg codecs for a given interchange lib from any task in the runtime. Right now we obviously only support `msgspec` but ideally this API holds general enough to be used for other backends eventually (like `capnproto`, and apache arrow). Impl is in a new `tractor.msg._codec` with: - a new `MsgCodec` type for encapsing `msgspec.msgpack.Encoder/Decoder` pairs and configuring any custom enc/dec_hooks or typed decoding. - factory `mk_codec()` for creating new codecs ad-hoc from a task. - `contextvars` support for a new `trio.Task` scoped `_ctxvar_MsgCodec: ContextVar[MsgCodec]` named 'msgspec_codec'. - `apply_codec()` for temporarily modifying the above per task as needed around `.open_context()` / `.open_stream()` operation. A new test (suite) in `test_caps_msging.py`: - verify a parent and its child can enable the same custom codec (in this case to transmit `NamespacePath`s) with tons of pedantic ctx-vars checks. - ToDo: still need to implement #36 msg types in order to be able to get decodes working (as in `MsgStream.receive()` will deliver an already created `NamespacePath` obj) since currently all msgs come packed in `dict`-msg wrapper packets.. -> use the proto from PR #35 to get nested `msgspec.Raw` processing up and running Bo
2024-03-26 19:50:47 +00:00
# NOTE: lookup the `trio.Task.context`'s var for
# the current `MsgCodec`.
2024-04-02 17:41:52 +00:00
codec: MsgCodec = _ctxvar_MsgCodec.get()
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
# XXX for ctxvar debug only!
# if self._codec.pld_spec != codec.pld_spec:
# assert (
# task := trio.lowlevel.current_task()
# ) is not self._task
# self._task = task
# self._codec = codec
# log.runtime(
# f'Using new codec in {self}.recv()\n'
# f'codec: {self._codec}\n\n'
# f'msg_bytes: {msg_bytes}\n'
# )
2024-04-02 17:41:52 +00:00
yield codec.decode(msg_bytes)
Proto `MsgCodec`, an interchange fmt modify API Fitting in line with the issues outstanding: - #36: (msg)spec-ing out our SCIPP (structured-con-inter-proc-prot). (https://github.com/goodboy/tractor/issues/36) - #196: adding strictly typed IPC msg dialog schemas, more or less better described as "dialog/transaction scoped message specs" using `msgspec`'s tagged unions and custom codecs. (https://github.com/goodboy/tractor/issues/196) - #365: using modern static type-annots to drive capability based messaging and RPC. (statically https://github.com/goodboy/tractor/issues/365) This is a first draft of a new API for dynamically overriding IPC msg codecs for a given interchange lib from any task in the runtime. Right now we obviously only support `msgspec` but ideally this API holds general enough to be used for other backends eventually (like `capnproto`, and apache arrow). Impl is in a new `tractor.msg._codec` with: - a new `MsgCodec` type for encapsing `msgspec.msgpack.Encoder/Decoder` pairs and configuring any custom enc/dec_hooks or typed decoding. - factory `mk_codec()` for creating new codecs ad-hoc from a task. - `contextvars` support for a new `trio.Task` scoped `_ctxvar_MsgCodec: ContextVar[MsgCodec]` named 'msgspec_codec'. - `apply_codec()` for temporarily modifying the above per task as needed around `.open_context()` / `.open_stream()` operation. A new test (suite) in `test_caps_msging.py`: - verify a parent and its child can enable the same custom codec (in this case to transmit `NamespacePath`s) with tons of pedantic ctx-vars checks. - ToDo: still need to implement #36 msg types in order to be able to get decodes working (as in `MsgStream.receive()` will deliver an already created `NamespacePath` obj) since currently all msgs come packed in `dict`-msg wrapper packets.. -> use the proto from PR #35 to get nested `msgspec.Raw` processing up and running Bo
2024-03-26 19:50:47 +00:00
2024-04-02 17:41:52 +00:00
# XXX NOTE: since the below error derives from
# `DecodeError` we need to catch is specially
# and always raise such that spec violations
# are never allowed to be caught silently!
except msgspec.ValidationError as verr:
msgtyperr: MsgTypeError = _mk_recv_mte(
msg=msg_bytes,
codec=codec,
Refine `MsgTypeError` handling to relay-up-on-`.recv()` Such that `Channel.recv()` + `MsgpackTCPStream.recv()` originating msg-type-errors are not raised at the IPC transport layer but instead relayed up the runtime stack for eventual handling by user-app code via the `Context`/`MsgStream` layer APIs. This design choice leads to a substantial amount of flexibility and modularity, and avoids `MsgTypeError` handling policies from being coupled to a particular backend IPC transport layer: - receive-side msg-type errors, as can be raised and handled in the `.open_stream()` "nasty" phase of a ctx, whilst being packed at the `MsgCodec`/transport layer (keeping the underlying src decode error coupled to the specific transport + interchange lib) and then relayed upward to app code for custom handling like a normal Error` msg. - the policy options for handling such cases could be implemented as `@acm` wrappers around `.open_context()`/`.open_stream()` blocks (and their respective delivered primitives) OR just plain old async generators around `MsgStream.receive()` such that both built-in policy handling and custom user-app solutions can be swapped without touching any `tractor` internals or providing specialized "registry APIs". -> eg. the ignore and relay-invalid-msg-to-sender approach can be more easily implemented as embedded `try: except MsgTypeError:` blocks around `MsgStream.receive()` possibly applied as either of an injected wrapper type around a stream or an async gen that `async for`s from the stream. - any performance based AOT-lang extensions used to implement a policy for handling recv-side errors space can avoid knowledge of the lower level IPC `Channel` (and-downward) primitives. - `Context` consuming code can choose to let all msg-type-errs bubble and handle them manually (like any other remote `Error` shuttled exception). - we can keep (as before) send-side msg type checks can be raised locally and cause offending senders to error and adjust before the streaming phase of an IPC ctx. Impl (related) deats: - obvi make `MsgpackTCPStream.recv()` yield up any `MsgTypeError` constructed by `_mk_msg_type_err()` such that the exception will eventually be relayed up to `._rpc.process_messages()` and from their delivered to the corresponding ctx-task. - in support of ^, make `Channel.recv()` detect said mtes and use the new `pack_from_raise()` to inject the far end `Actor.uid` for the `Error.src_uid`. - keep raising the send side equivalent (when strict enabled) errors inline immediately with no upward `Error` packing or relay. - improve `_mk_msg_type_err()` cases handling with far more detailed `MsgTypeError` "message" contents pertaining to `msgspec` specific failure-fixing-tips and type-spec mismatch info: * use `.from_decode()` constructor in recv-side case to inject the non-spec decoded `msg_dict: dict` and use the new `MsgCodec.pld_spec_str: str` when clarifying the type discrepancy with the offending field. * on send-side, if we detect that an unsupported field type was described in the original `src_type_error`, AND there is no `msgpack.Encoder.enc_hook()` set, that the real issue is likely that the user needs to extend the codec to support the non-std/custom type with a hook and link to `msgspec` docs. * if one of a `src_type/validation_error` is provided, set that error as the `.__cause__` in the new mte.
2024-04-09 14:36:25 +00:00
src_validation_error=verr,
2024-04-02 17:41:52 +00:00
)
Refine `MsgTypeError` handling to relay-up-on-`.recv()` Such that `Channel.recv()` + `MsgpackTCPStream.recv()` originating msg-type-errors are not raised at the IPC transport layer but instead relayed up the runtime stack for eventual handling by user-app code via the `Context`/`MsgStream` layer APIs. This design choice leads to a substantial amount of flexibility and modularity, and avoids `MsgTypeError` handling policies from being coupled to a particular backend IPC transport layer: - receive-side msg-type errors, as can be raised and handled in the `.open_stream()` "nasty" phase of a ctx, whilst being packed at the `MsgCodec`/transport layer (keeping the underlying src decode error coupled to the specific transport + interchange lib) and then relayed upward to app code for custom handling like a normal Error` msg. - the policy options for handling such cases could be implemented as `@acm` wrappers around `.open_context()`/`.open_stream()` blocks (and their respective delivered primitives) OR just plain old async generators around `MsgStream.receive()` such that both built-in policy handling and custom user-app solutions can be swapped without touching any `tractor` internals or providing specialized "registry APIs". -> eg. the ignore and relay-invalid-msg-to-sender approach can be more easily implemented as embedded `try: except MsgTypeError:` blocks around `MsgStream.receive()` possibly applied as either of an injected wrapper type around a stream or an async gen that `async for`s from the stream. - any performance based AOT-lang extensions used to implement a policy for handling recv-side errors space can avoid knowledge of the lower level IPC `Channel` (and-downward) primitives. - `Context` consuming code can choose to let all msg-type-errs bubble and handle them manually (like any other remote `Error` shuttled exception). - we can keep (as before) send-side msg type checks can be raised locally and cause offending senders to error and adjust before the streaming phase of an IPC ctx. Impl (related) deats: - obvi make `MsgpackTCPStream.recv()` yield up any `MsgTypeError` constructed by `_mk_msg_type_err()` such that the exception will eventually be relayed up to `._rpc.process_messages()` and from their delivered to the corresponding ctx-task. - in support of ^, make `Channel.recv()` detect said mtes and use the new `pack_from_raise()` to inject the far end `Actor.uid` for the `Error.src_uid`. - keep raising the send side equivalent (when strict enabled) errors inline immediately with no upward `Error` packing or relay. - improve `_mk_msg_type_err()` cases handling with far more detailed `MsgTypeError` "message" contents pertaining to `msgspec` specific failure-fixing-tips and type-spec mismatch info: * use `.from_decode()` constructor in recv-side case to inject the non-spec decoded `msg_dict: dict` and use the new `MsgCodec.pld_spec_str: str` when clarifying the type discrepancy with the offending field. * on send-side, if we detect that an unsupported field type was described in the original `src_type_error`, AND there is no `msgpack.Encoder.enc_hook()` set, that the real issue is likely that the user needs to extend the codec to support the non-std/custom type with a hook and link to `msgspec` docs. * if one of a `src_type/validation_error` is provided, set that error as the `.__cause__` in the new mte.
2024-04-09 14:36:25 +00:00
# XXX deliver up to `Channel.recv()` where
# a re-raise and `Error`-pack can inject the far
# end actor `.uid`.
yield msgtyperr
2024-04-02 17:41:52 +00:00
except (
2022-02-16 18:06:24 +00:00
msgspec.DecodeError,
UnicodeDecodeError,
):
if decodes_failed < 4:
# ignore decoding errors for now and assume they have to
# do with a channel drop - hope that receiving from the
# channel will raise an expected error and bubble up.
2022-07-01 18:37:46 +00:00
try:
2024-04-02 17:41:52 +00:00
msg_str: str|bytes = msg_bytes.decode()
2022-07-01 18:37:46 +00:00
except UnicodeDecodeError:
msg_str = msg_bytes
2022-07-01 18:37:46 +00:00
2024-04-02 17:41:52 +00:00
log.exception(
'Failed to decode msg?\n'
f'{codec}\n\n'
'Rxed bytes from wire:\n\n'
f'{msg_str!r}\n'
)
decodes_failed += 1
else:
raise
async def send(
self,
msg: msgtypes.MsgType,
strict_types: bool = True,
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
hide_tb: bool = False,
) -> None:
'''
Send a msgpack encoded py-object-blob-as-msg over TCP.
If `strict_types == True` then a `MsgTypeError` will be raised on any
invalid msg type
'''
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
__tracebackhide__: bool = hide_tb
# XXX see `trio._sync.AsyncContextManagerMixin` for details
# on the `.acquire()`/`.release()` sequencing..
async with self._send_lock:
Proto `MsgCodec`, an interchange fmt modify API Fitting in line with the issues outstanding: - #36: (msg)spec-ing out our SCIPP (structured-con-inter-proc-prot). (https://github.com/goodboy/tractor/issues/36) - #196: adding strictly typed IPC msg dialog schemas, more or less better described as "dialog/transaction scoped message specs" using `msgspec`'s tagged unions and custom codecs. (https://github.com/goodboy/tractor/issues/196) - #365: using modern static type-annots to drive capability based messaging and RPC. (statically https://github.com/goodboy/tractor/issues/365) This is a first draft of a new API for dynamically overriding IPC msg codecs for a given interchange lib from any task in the runtime. Right now we obviously only support `msgspec` but ideally this API holds general enough to be used for other backends eventually (like `capnproto`, and apache arrow). Impl is in a new `tractor.msg._codec` with: - a new `MsgCodec` type for encapsing `msgspec.msgpack.Encoder/Decoder` pairs and configuring any custom enc/dec_hooks or typed decoding. - factory `mk_codec()` for creating new codecs ad-hoc from a task. - `contextvars` support for a new `trio.Task` scoped `_ctxvar_MsgCodec: ContextVar[MsgCodec]` named 'msgspec_codec'. - `apply_codec()` for temporarily modifying the above per task as needed around `.open_context()` / `.open_stream()` operation. A new test (suite) in `test_caps_msging.py`: - verify a parent and its child can enable the same custom codec (in this case to transmit `NamespacePath`s) with tons of pedantic ctx-vars checks. - ToDo: still need to implement #36 msg types in order to be able to get decodes working (as in `MsgStream.receive()` will deliver an already created `NamespacePath` obj) since currently all msgs come packed in `dict`-msg wrapper packets.. -> use the proto from PR #35 to get nested `msgspec.Raw` processing up and running Bo
2024-03-26 19:50:47 +00:00
# NOTE: lookup the `trio.Task.context`'s var for
# the current `MsgCodec`.
2024-04-02 17:41:52 +00:00
codec: MsgCodec = _ctxvar_MsgCodec.get()
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
# XXX for ctxvar debug only!
# if self._codec.pld_spec != codec.pld_spec:
# self._codec = codec
# log.runtime(
# f'Using new codec in {self}.send()\n'
# f'codec: {self._codec}\n\n'
# f'msg: {msg}\n'
# )
if type(msg) not in msgtypes.__msg_types__:
if strict_types:
raise _mk_send_mte(
msg,
codec=codec,
)
else:
log.warning(
'Sending non-`Msg`-spec msg?\n\n'
f'{msg}\n'
)
try:
bytes_data: bytes = codec.encode(msg)
except TypeError as _err:
typerr = _err
msgtyperr: MsgTypeError = _mk_send_mte(
Refine `MsgTypeError` handling to relay-up-on-`.recv()` Such that `Channel.recv()` + `MsgpackTCPStream.recv()` originating msg-type-errors are not raised at the IPC transport layer but instead relayed up the runtime stack for eventual handling by user-app code via the `Context`/`MsgStream` layer APIs. This design choice leads to a substantial amount of flexibility and modularity, and avoids `MsgTypeError` handling policies from being coupled to a particular backend IPC transport layer: - receive-side msg-type errors, as can be raised and handled in the `.open_stream()` "nasty" phase of a ctx, whilst being packed at the `MsgCodec`/transport layer (keeping the underlying src decode error coupled to the specific transport + interchange lib) and then relayed upward to app code for custom handling like a normal Error` msg. - the policy options for handling such cases could be implemented as `@acm` wrappers around `.open_context()`/`.open_stream()` blocks (and their respective delivered primitives) OR just plain old async generators around `MsgStream.receive()` such that both built-in policy handling and custom user-app solutions can be swapped without touching any `tractor` internals or providing specialized "registry APIs". -> eg. the ignore and relay-invalid-msg-to-sender approach can be more easily implemented as embedded `try: except MsgTypeError:` blocks around `MsgStream.receive()` possibly applied as either of an injected wrapper type around a stream or an async gen that `async for`s from the stream. - any performance based AOT-lang extensions used to implement a policy for handling recv-side errors space can avoid knowledge of the lower level IPC `Channel` (and-downward) primitives. - `Context` consuming code can choose to let all msg-type-errs bubble and handle them manually (like any other remote `Error` shuttled exception). - we can keep (as before) send-side msg type checks can be raised locally and cause offending senders to error and adjust before the streaming phase of an IPC ctx. Impl (related) deats: - obvi make `MsgpackTCPStream.recv()` yield up any `MsgTypeError` constructed by `_mk_msg_type_err()` such that the exception will eventually be relayed up to `._rpc.process_messages()` and from their delivered to the corresponding ctx-task. - in support of ^, make `Channel.recv()` detect said mtes and use the new `pack_from_raise()` to inject the far end `Actor.uid` for the `Error.src_uid`. - keep raising the send side equivalent (when strict enabled) errors inline immediately with no upward `Error` packing or relay. - improve `_mk_msg_type_err()` cases handling with far more detailed `MsgTypeError` "message" contents pertaining to `msgspec` specific failure-fixing-tips and type-spec mismatch info: * use `.from_decode()` constructor in recv-side case to inject the non-spec decoded `msg_dict: dict` and use the new `MsgCodec.pld_spec_str: str` when clarifying the type discrepancy with the offending field. * on send-side, if we detect that an unsupported field type was described in the original `src_type_error`, AND there is no `msgpack.Encoder.enc_hook()` set, that the real issue is likely that the user needs to extend the codec to support the non-std/custom type with a hook and link to `msgspec` docs. * if one of a `src_type/validation_error` is provided, set that error as the `.__cause__` in the new mte.
2024-04-09 14:36:25 +00:00
msg,
codec=codec,
message=(
f'IPC-msg-spec violation in\n\n'
f'{pretty_struct.Struct.pformat(msg)}'
),
src_type_error=typerr,
)
raise msgtyperr from typerr
# supposedly the fastest says,
# https://stackoverflow.com/a/54027962
2021-07-01 18:52:52 +00:00
size: bytes = struct.pack("<I", len(bytes_data))
return await self.stream.send_all(size + bytes_data)
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
# ?TODO? does it help ever to dynamically show this
# frame?
# try:
# <the-above_code>
# except BaseException as _err:
# err = _err
# if not isinstance(err, MsgTypeError):
# __tracebackhide__: bool = False
# raise
@property
2022-07-12 15:22:30 +00:00
def laddr(self) -> tuple[str, int]:
return self._laddr
@property
2022-07-12 15:22:30 +00:00
def raddr(self) -> tuple[str, int]:
return self._raddr
async def recv(self) -> Any:
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
return await self._aiter_pkts.asend(None)
async def drain(self) -> AsyncIterator[dict]:
'''
Drain the stream's remaining messages sent from
the far end until the connection is closed by
the peer.
'''
try:
async for msg in self._iter_packets():
self.drained.append(msg)
except TransportClosed:
for msg in self.drained:
yield msg
def __aiter__(self):
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
return self._aiter_pkts
def connected(self) -> bool:
return self.stream.socket.fileno() != -1
def get_msg_transport(
2022-07-12 15:22:30 +00:00
key: tuple[str, str],
) -> Type[MsgTransport]:
2021-09-08 01:07:33 +00:00
return {
('msgpack', 'tcp'): MsgpackTCPStream,
}[key]
2021-09-08 01:07:33 +00:00
class Channel:
'''
An inter-process channel for communication between (remote) actors.
2018-05-30 16:36:23 +00:00
Wraps a ``MsgStream``: transport + encoding IPC connection.
2022-07-12 15:22:30 +00:00
Currently we only support ``trio.SocketStream`` for transport
2022-07-12 15:22:30 +00:00
(aka TCP) and the ``msgpack`` interchange format via the ``msgspec``
codec libary.
2021-09-08 01:07:33 +00:00
'''
2018-05-30 16:36:23 +00:00
def __init__(
self,
destaddr: tuple[str, int]|None,
2022-07-12 15:22:30 +00:00
msg_transport_type_key: tuple[str, str] = ('msgpack', 'tcp'),
# TODO: optional reconnection support?
# auto_reconnect: bool = False,
# on_reconnect: typing.Callable[..., typing.Awaitable] = None,
) -> None:
# self._recon_seq = on_reconnect
# self._autorecon = auto_reconnect
self._destaddr = destaddr
self._transport_key = msg_transport_type_key
# Either created in ``.connect()`` or passed in by
# user in ``.from_stream()``.
self._stream: trio.SocketStream|None = None
self._transport: MsgTransport|None = None
2021-07-01 18:52:52 +00:00
2018-06-21 21:09:22 +00:00
# set after handshake - always uid of far end
self.uid: tuple[str, str]|None = None
2021-07-01 18:52:52 +00:00
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
self._aiter_msgs = self._iter_msgs()
self._exc: Exception|None = None # set if far end actor errors
self._closed: bool = False
2023-10-23 21:47:55 +00:00
# flag set by ``Portal.cancel_actor()`` indicating remote
# (possibly peer) cancellation of the far end actor
# runtime.
self._cancel_called: bool = False
@property
def msgstream(self) -> MsgTransport:
Proto `MsgCodec`, an interchange fmt modify API Fitting in line with the issues outstanding: - #36: (msg)spec-ing out our SCIPP (structured-con-inter-proc-prot). (https://github.com/goodboy/tractor/issues/36) - #196: adding strictly typed IPC msg dialog schemas, more or less better described as "dialog/transaction scoped message specs" using `msgspec`'s tagged unions and custom codecs. (https://github.com/goodboy/tractor/issues/196) - #365: using modern static type-annots to drive capability based messaging and RPC. (statically https://github.com/goodboy/tractor/issues/365) This is a first draft of a new API for dynamically overriding IPC msg codecs for a given interchange lib from any task in the runtime. Right now we obviously only support `msgspec` but ideally this API holds general enough to be used for other backends eventually (like `capnproto`, and apache arrow). Impl is in a new `tractor.msg._codec` with: - a new `MsgCodec` type for encapsing `msgspec.msgpack.Encoder/Decoder` pairs and configuring any custom enc/dec_hooks or typed decoding. - factory `mk_codec()` for creating new codecs ad-hoc from a task. - `contextvars` support for a new `trio.Task` scoped `_ctxvar_MsgCodec: ContextVar[MsgCodec]` named 'msgspec_codec'. - `apply_codec()` for temporarily modifying the above per task as needed around `.open_context()` / `.open_stream()` operation. A new test (suite) in `test_caps_msging.py`: - verify a parent and its child can enable the same custom codec (in this case to transmit `NamespacePath`s) with tons of pedantic ctx-vars checks. - ToDo: still need to implement #36 msg types in order to be able to get decodes working (as in `MsgStream.receive()` will deliver an already created `NamespacePath` obj) since currently all msgs come packed in `dict`-msg wrapper packets.. -> use the proto from PR #35 to get nested `msgspec.Raw` processing up and running Bo
2024-03-26 19:50:47 +00:00
log.info(
'`Channel.msgstream` is an old name, use `._transport`'
)
return self._transport
@property
def transport(self) -> MsgTransport:
return self._transport
@classmethod
def from_stream(
cls,
stream: trio.SocketStream,
**kwargs,
) -> Channel:
src, dst = get_stream_addrs(stream)
chan = Channel(
destaddr=dst,
**kwargs,
)
# set immediately here from provided instance
chan._stream: trio.SocketStream = stream
chan.set_msg_transport(stream)
return chan
def set_msg_transport(
self,
stream: trio.SocketStream,
type_key: tuple[str, str]|None = None,
# XXX optionally provided codec pair for `msgspec`:
# https://jcristharif.com/msgspec/extending.html#mapping-to-from-native-types
Proto `MsgCodec`, an interchange fmt modify API Fitting in line with the issues outstanding: - #36: (msg)spec-ing out our SCIPP (structured-con-inter-proc-prot). (https://github.com/goodboy/tractor/issues/36) - #196: adding strictly typed IPC msg dialog schemas, more or less better described as "dialog/transaction scoped message specs" using `msgspec`'s tagged unions and custom codecs. (https://github.com/goodboy/tractor/issues/196) - #365: using modern static type-annots to drive capability based messaging and RPC. (statically https://github.com/goodboy/tractor/issues/365) This is a first draft of a new API for dynamically overriding IPC msg codecs for a given interchange lib from any task in the runtime. Right now we obviously only support `msgspec` but ideally this API holds general enough to be used for other backends eventually (like `capnproto`, and apache arrow). Impl is in a new `tractor.msg._codec` with: - a new `MsgCodec` type for encapsing `msgspec.msgpack.Encoder/Decoder` pairs and configuring any custom enc/dec_hooks or typed decoding. - factory `mk_codec()` for creating new codecs ad-hoc from a task. - `contextvars` support for a new `trio.Task` scoped `_ctxvar_MsgCodec: ContextVar[MsgCodec]` named 'msgspec_codec'. - `apply_codec()` for temporarily modifying the above per task as needed around `.open_context()` / `.open_stream()` operation. A new test (suite) in `test_caps_msging.py`: - verify a parent and its child can enable the same custom codec (in this case to transmit `NamespacePath`s) with tons of pedantic ctx-vars checks. - ToDo: still need to implement #36 msg types in order to be able to get decodes working (as in `MsgStream.receive()` will deliver an already created `NamespacePath` obj) since currently all msgs come packed in `dict`-msg wrapper packets.. -> use the proto from PR #35 to get nested `msgspec.Raw` processing up and running Bo
2024-03-26 19:50:47 +00:00
codec: MsgCodec|None = None,
) -> MsgTransport:
type_key = (
type_key
or
self._transport_key
)
Proto `MsgCodec`, an interchange fmt modify API Fitting in line with the issues outstanding: - #36: (msg)spec-ing out our SCIPP (structured-con-inter-proc-prot). (https://github.com/goodboy/tractor/issues/36) - #196: adding strictly typed IPC msg dialog schemas, more or less better described as "dialog/transaction scoped message specs" using `msgspec`'s tagged unions and custom codecs. (https://github.com/goodboy/tractor/issues/196) - #365: using modern static type-annots to drive capability based messaging and RPC. (statically https://github.com/goodboy/tractor/issues/365) This is a first draft of a new API for dynamically overriding IPC msg codecs for a given interchange lib from any task in the runtime. Right now we obviously only support `msgspec` but ideally this API holds general enough to be used for other backends eventually (like `capnproto`, and apache arrow). Impl is in a new `tractor.msg._codec` with: - a new `MsgCodec` type for encapsing `msgspec.msgpack.Encoder/Decoder` pairs and configuring any custom enc/dec_hooks or typed decoding. - factory `mk_codec()` for creating new codecs ad-hoc from a task. - `contextvars` support for a new `trio.Task` scoped `_ctxvar_MsgCodec: ContextVar[MsgCodec]` named 'msgspec_codec'. - `apply_codec()` for temporarily modifying the above per task as needed around `.open_context()` / `.open_stream()` operation. A new test (suite) in `test_caps_msging.py`: - verify a parent and its child can enable the same custom codec (in this case to transmit `NamespacePath`s) with tons of pedantic ctx-vars checks. - ToDo: still need to implement #36 msg types in order to be able to get decodes working (as in `MsgStream.receive()` will deliver an already created `NamespacePath` obj) since currently all msgs come packed in `dict`-msg wrapper packets.. -> use the proto from PR #35 to get nested `msgspec.Raw` processing up and running Bo
2024-03-26 19:50:47 +00:00
# get transport type, then
self._transport = get_msg_transport(
type_key
Proto `MsgCodec`, an interchange fmt modify API Fitting in line with the issues outstanding: - #36: (msg)spec-ing out our SCIPP (structured-con-inter-proc-prot). (https://github.com/goodboy/tractor/issues/36) - #196: adding strictly typed IPC msg dialog schemas, more or less better described as "dialog/transaction scoped message specs" using `msgspec`'s tagged unions and custom codecs. (https://github.com/goodboy/tractor/issues/196) - #365: using modern static type-annots to drive capability based messaging and RPC. (statically https://github.com/goodboy/tractor/issues/365) This is a first draft of a new API for dynamically overriding IPC msg codecs for a given interchange lib from any task in the runtime. Right now we obviously only support `msgspec` but ideally this API holds general enough to be used for other backends eventually (like `capnproto`, and apache arrow). Impl is in a new `tractor.msg._codec` with: - a new `MsgCodec` type for encapsing `msgspec.msgpack.Encoder/Decoder` pairs and configuring any custom enc/dec_hooks or typed decoding. - factory `mk_codec()` for creating new codecs ad-hoc from a task. - `contextvars` support for a new `trio.Task` scoped `_ctxvar_MsgCodec: ContextVar[MsgCodec]` named 'msgspec_codec'. - `apply_codec()` for temporarily modifying the above per task as needed around `.open_context()` / `.open_stream()` operation. A new test (suite) in `test_caps_msging.py`: - verify a parent and its child can enable the same custom codec (in this case to transmit `NamespacePath`s) with tons of pedantic ctx-vars checks. - ToDo: still need to implement #36 msg types in order to be able to get decodes working (as in `MsgStream.receive()` will deliver an already created `NamespacePath` obj) since currently all msgs come packed in `dict`-msg wrapper packets.. -> use the proto from PR #35 to get nested `msgspec.Raw` processing up and running Bo
2024-03-26 19:50:47 +00:00
# instantiate an instance of the msg-transport
)(
stream,
codec=codec,
)
return self._transport
Proto `MsgCodec`, an interchange fmt modify API Fitting in line with the issues outstanding: - #36: (msg)spec-ing out our SCIPP (structured-con-inter-proc-prot). (https://github.com/goodboy/tractor/issues/36) - #196: adding strictly typed IPC msg dialog schemas, more or less better described as "dialog/transaction scoped message specs" using `msgspec`'s tagged unions and custom codecs. (https://github.com/goodboy/tractor/issues/196) - #365: using modern static type-annots to drive capability based messaging and RPC. (statically https://github.com/goodboy/tractor/issues/365) This is a first draft of a new API for dynamically overriding IPC msg codecs for a given interchange lib from any task in the runtime. Right now we obviously only support `msgspec` but ideally this API holds general enough to be used for other backends eventually (like `capnproto`, and apache arrow). Impl is in a new `tractor.msg._codec` with: - a new `MsgCodec` type for encapsing `msgspec.msgpack.Encoder/Decoder` pairs and configuring any custom enc/dec_hooks or typed decoding. - factory `mk_codec()` for creating new codecs ad-hoc from a task. - `contextvars` support for a new `trio.Task` scoped `_ctxvar_MsgCodec: ContextVar[MsgCodec]` named 'msgspec_codec'. - `apply_codec()` for temporarily modifying the above per task as needed around `.open_context()` / `.open_stream()` operation. A new test (suite) in `test_caps_msging.py`: - verify a parent and its child can enable the same custom codec (in this case to transmit `NamespacePath`s) with tons of pedantic ctx-vars checks. - ToDo: still need to implement #36 msg types in order to be able to get decodes working (as in `MsgStream.receive()` will deliver an already created `NamespacePath` obj) since currently all msgs come packed in `dict`-msg wrapper packets.. -> use the proto from PR #35 to get nested `msgspec.Raw` processing up and running Bo
2024-03-26 19:50:47 +00:00
@cm
def apply_codec(
self,
codec: MsgCodec,
) -> None:
'''
Temporarily override the underlying IPC msg codec for
dynamic enforcement of messaging schema.
'''
orig: MsgCodec = self._transport.codec
try:
self._transport.codec = codec
yield
finally:
self._transport.codec = orig
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
# TODO: do a .src/.dst: str for maddrs?
def __repr__(self) -> str:
if not self._transport:
return '<Channel with inactive transport?>'
return repr(
self._transport.stream.socket._sock
).replace( # type: ignore
"socket.socket",
"Channel",
)
@property
def laddr(self) -> tuple[str, int]|None:
return self._transport.laddr if self._transport else None
@property
def raddr(self) -> tuple[str, int]|None:
return self._transport.raddr if self._transport else None
async def connect(
self,
destaddr: tuple[Any, ...] | None = None,
2019-12-10 05:55:03 +00:00
**kwargs
) -> MsgTransport:
2018-07-04 07:16:00 +00:00
if self.connected():
raise RuntimeError("channel is already connected?")
destaddr = destaddr or self._destaddr
2019-12-10 05:55:03 +00:00
assert isinstance(destaddr, tuple)
2021-06-30 17:47:07 +00:00
stream = await trio.open_tcp_stream(
*destaddr,
**kwargs
)
transport = self.set_msg_transport(stream)
2021-06-30 17:47:07 +00:00
log.transport(
f'Opened channel[{type(transport)}]: {self.laddr} -> {self.raddr}'
2021-06-30 17:47:07 +00:00
)
return transport
2018-05-30 16:36:23 +00:00
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
# TODO: something like,
# `pdbp.hideframe_on(errors=[MsgTypeError])`
# instead of the `try/except` hack we have rn..
# seems like a pretty useful thing to have in general
# along with being able to filter certain stack frame(s / sets)
# possibly based on the current log-level?
async def send(
self,
payload: Any,
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
hide_tb: bool = False,
) -> None:
'''
Send a coded msg-blob over the transport.
'''
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
__tracebackhide__: bool = hide_tb
try:
log.transport(
'=> send IPC msg:\n\n'
f'{pformat(payload)}\n'
)
# assert self._transport # but why typing?
await self._transport.send(
payload,
hide_tb=hide_tb,
)
except BaseException as _err:
err = _err # bind for introspection
if not isinstance(_err, MsgTypeError):
# assert err
__tracebackhide__: bool = False
else:
assert err.cid
raise
2018-05-30 16:36:23 +00:00
2018-08-31 21:16:24 +00:00
async def recv(self) -> Any:
assert self._transport
return await self._transport.recv()
# TODO: auto-reconnect features like 0mq/nanomsg?
# -[ ] implement it manually with nods to SC prot
# possibly on multiple transport backends?
# -> seems like that might be re-inventing scalability
# prots tho no?
# try:
# return await self._transport.recv()
# except trio.BrokenResourceError:
# if self._autorecon:
# await self._reconnect()
# return await self.recv()
# raise
async def aclose(self) -> None:
2021-06-30 17:47:07 +00:00
log.transport(
f'Closing channel to {self.uid} '
f'{self.laddr} -> {self.raddr}'
)
assert self._transport
await self._transport.stream.aclose()
self._closed = True
2018-05-30 16:36:23 +00:00
async def __aenter__(self):
await self.connect()
2018-05-30 16:36:23 +00:00
return self
async def __aexit__(self, *args):
await self.aclose(*args)
def __aiter__(self):
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
return self._aiter_msgs
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
# ?TODO? run any reconnection sequence?
# -[ ] prolly should be impl-ed as deco-API?
#
# async def _reconnect(self) -> None:
# """Handle connection failures by polling until a reconnect can be
# established.
# """
# down = False
# while True:
# try:
# with trio.move_on_after(3) as cancel_scope:
# await self.connect()
# cancelled = cancel_scope.cancelled_caught
# if cancelled:
# log.transport(
# "Reconnect timed out after 3 seconds, retrying...")
# continue
# else:
# log.transport("Stream connection re-established!")
# # on_recon = self._recon_seq
# # if on_recon:
# # await on_recon(self)
# break
# except (OSError, ConnectionRefusedError):
# if not down:
# down = True
# log.transport(
# f"Connection to {self.raddr} went down, waiting"
# " for re-establishment")
# await trio.sleep(1)
2018-05-30 16:36:23 +00:00
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
async def _iter_msgs(
self
2021-12-02 17:34:27 +00:00
) -> AsyncGenerator[Any, None]:
'''
More formal `TransportClosed` reporting/raising Since it was all ad-hoc defined inside `._ipc.MsgpackTCPStream._iter_pkts()` more or less, this starts formalizing a way for particular transport backends to indicate whether a disconnect condition should be re-raised in the RPC msg loop and if not what log level to report it at (if any). Based on our lone transport currently we try to suppress any logging noise from ephemeral connections expected during normal actor interaction and discovery subsys ops: - any short lived discovery related TCP connects are only logged as `.transport()` level. - both `.error()` and raise on any underlying `trio.ClosedResource` cause since that normally means some task touched transport layer internals that it shouldn't have. - do a `.warning()` on anything else unexpected. Impl deats: - extend the `._exceptions.TransportClosed` to accept an input log level, raise-on-report toggle and custom reporting & raising via a new `.report_n_maybe_raise()` method. - construct the TCs with inputs per case in (the newly named) `._iter_pkts(). - call ^ this method from the `TransportClosed` handler block inside the RPC msg loop thus delegating reporting levels and/or raising to the backend's per-case TC instantiating. Related `._ipc` changes: - mask out all the `MsgpackTCPStream._codec` debug helper stuff and drop any lingering cruft from the initial proto-ing of msg-codecs. - rename some attrs/methods: |_`MsgpackTCPStream._iter_packets()` -> `._iter_pkts()` and `._agen` -> `_aiter_pkts`. |_`Channel._aiter_recv()` -> `._aiter_msgs()` and `._agen` -> `_aiter_msgs`. - add `hide_tb: bool` support to `Channel.send()` and only show the frame on non-MTEs.
2024-07-02 16:21:26 +00:00
Yield `MsgType` IPC msgs decoded and deliverd from
an underlying `MsgTransport` protocol.
This is a streaming routine alo implemented as an async-gen
func (same a `MsgTransport._iter_pkts()`) gets allocated by
a `.__call__()` inside `.__init__()` where it is assigned to
the `._aiter_msgs` attr.
2021-12-02 17:34:27 +00:00
'''
assert self._transport
2018-05-30 16:36:23 +00:00
while True:
try:
Refine `MsgTypeError` handling to relay-up-on-`.recv()` Such that `Channel.recv()` + `MsgpackTCPStream.recv()` originating msg-type-errors are not raised at the IPC transport layer but instead relayed up the runtime stack for eventual handling by user-app code via the `Context`/`MsgStream` layer APIs. This design choice leads to a substantial amount of flexibility and modularity, and avoids `MsgTypeError` handling policies from being coupled to a particular backend IPC transport layer: - receive-side msg-type errors, as can be raised and handled in the `.open_stream()` "nasty" phase of a ctx, whilst being packed at the `MsgCodec`/transport layer (keeping the underlying src decode error coupled to the specific transport + interchange lib) and then relayed upward to app code for custom handling like a normal Error` msg. - the policy options for handling such cases could be implemented as `@acm` wrappers around `.open_context()`/`.open_stream()` blocks (and their respective delivered primitives) OR just plain old async generators around `MsgStream.receive()` such that both built-in policy handling and custom user-app solutions can be swapped without touching any `tractor` internals or providing specialized "registry APIs". -> eg. the ignore and relay-invalid-msg-to-sender approach can be more easily implemented as embedded `try: except MsgTypeError:` blocks around `MsgStream.receive()` possibly applied as either of an injected wrapper type around a stream or an async gen that `async for`s from the stream. - any performance based AOT-lang extensions used to implement a policy for handling recv-side errors space can avoid knowledge of the lower level IPC `Channel` (and-downward) primitives. - `Context` consuming code can choose to let all msg-type-errs bubble and handle them manually (like any other remote `Error` shuttled exception). - we can keep (as before) send-side msg type checks can be raised locally and cause offending senders to error and adjust before the streaming phase of an IPC ctx. Impl (related) deats: - obvi make `MsgpackTCPStream.recv()` yield up any `MsgTypeError` constructed by `_mk_msg_type_err()` such that the exception will eventually be relayed up to `._rpc.process_messages()` and from their delivered to the corresponding ctx-task. - in support of ^, make `Channel.recv()` detect said mtes and use the new `pack_from_raise()` to inject the far end `Actor.uid` for the `Error.src_uid`. - keep raising the send side equivalent (when strict enabled) errors inline immediately with no upward `Error` packing or relay. - improve `_mk_msg_type_err()` cases handling with far more detailed `MsgTypeError` "message" contents pertaining to `msgspec` specific failure-fixing-tips and type-spec mismatch info: * use `.from_decode()` constructor in recv-side case to inject the non-spec decoded `msg_dict: dict` and use the new `MsgCodec.pld_spec_str: str` when clarifying the type discrepancy with the offending field. * on send-side, if we detect that an unsupported field type was described in the original `src_type_error`, AND there is no `msgpack.Encoder.enc_hook()` set, that the real issue is likely that the user needs to extend the codec to support the non-std/custom type with a hook and link to `msgspec` docs. * if one of a `src_type/validation_error` is provided, set that error as the `.__cause__` in the new mte.
2024-04-09 14:36:25 +00:00
async for msg in self._transport:
match msg:
# NOTE: if transport/interchange delivers
# a type error, we pack it with the far
# end peer `Actor.uid` and relay the
# `Error`-msg upward to the `._rpc` stack
# for normal RAE handling.
case MsgTypeError():
yield pack_from_raise(
local_err=msg,
cid=msg.cid,
# XXX we pack it here bc lower
# layers have no notion of an
# actor-id ;)
src_uid=self.uid,
)
case _:
yield msg
2018-11-09 06:53:15 +00:00
except trio.BrokenResourceError:
# if not self._autorecon:
raise
2018-07-04 07:16:00 +00:00
await self.aclose()
# if self._autorecon: # attempt reconnect
# await self._reconnect()
# continue
2018-06-21 21:09:22 +00:00
def connected(self) -> bool:
return self._transport.connected() if self._transport else False
2018-07-11 22:08:57 +00:00
@acm
async def _connect_chan(
Init-support for "multi homed" transports Since we'd like to eventually allow a diverse set of transport (protocol) methods and stacks, and a multi-peer discovery system for distributed actor-tree applications, this reworks all runtime internals to support multi-homing for any given tree on a logical host. In other words any actor can now bind its transport server (currently only unsecured TCP + `msgspec`) to more then one address available in its (linux) network namespace. Further, registry actors (now dubbed "registars" instead of "arbiters") can also similarly bind to multiple network addresses and provide discovery services to remote actors via multiple addresses which can now be provided at runtime startup. Deats: - adjust `._runtime` internals to use a `list[tuple[str, int]]` (and thus pluralized) socket address sequence where applicable for transport server socket binds, now exposed via `Actor.accept_addrs`: - `Actor.__init__()` now takes a `registry_addrs: list`. - `Actor.is_arbiter` -> `.is_registrar`. - `._arb_addr` -> `._reg_addrs: list[tuple]`. - always reg and de-reg from all registrars in `async_main()`. - only set the global runtime var `'_root_mailbox'` to the loopback address since normally all in-tree processes should have access to it, right? - `._serve_forever()` task now takes `listen_sockaddrs: list[tuple]` - make `open_root_actor()` take a `registry_addrs: list[tuple[str, int]]` and defaults when not passed. - change `ActorNursery.start_..()` methods take `bind_addrs: list` and pass down through the spawning layer(s) via the parent-seed-msg. - generalize all `._discovery()` APIs to accept `registry_addrs`-like inputs and move all relevant subsystems to adopt the "registry" style naming instead of "arbiter": - make `find_actor()` support batched concurrent portal queries over all provided input addresses using `.trionics.gather_contexts()` Bo - syntax: move to using `async with <tuples>` 3.9+ style chained @acms. - a general modernization of the code to a python 3.9+ style. - start deprecation and change to "registry" naming / semantics: - `._discovery.get_arbiter()` -> `.get_registry()`
2023-09-27 19:19:30 +00:00
host: str,
port: int
2018-08-31 21:16:24 +00:00
) -> typing.AsyncGenerator[Channel, None]:
2021-12-02 17:34:27 +00:00
'''
Create and connect a channel with disconnect on context manager
teardown.
2021-12-02 17:34:27 +00:00
'''
2018-07-11 22:08:57 +00:00
chan = Channel((host, port))
await chan.connect()
yield chan
with trio.CancelScope(shield=True):
await chan.aclose()