1.4.0

What's Changed

Essentially, all that changed is that ConnectionParameters.from_pydantic_multihost_hosts() now accepts hosts from multi-host url that has one credentials pair, like this: ArtemisDsn("tcp://lev:pass@host1:61616,host2:61617,host3:61618").hosts(). Previously you had to specify credentials for each host: ArtemisDsn("tcp://user1:pass1@host1:61616,user2:pass2@host2:61617,user3:pass@host3:61618").hosts().

---

• Expect only one username-password pair in ConnectionParameters.from_pydantic_multihost_hosts() by @vrslev in #72
• Rename ConnectionParameters.from_pydantic_multihost_hosts() to ConnectionParameters.from_pydantic_multihosturl_hosts() by @vrslev in #73
• Fix tests by @vrslev in #74
• Revert recent change by @vrslev in #75
• Allow to set-credentials to all hosts (old style) by @vrslev in #76

Full Changelog: 1.3.1...1.4.0

1.3.1

What's Changed

• Add notice about SSL to readme by @vrslev in #67
• Fix SSL by @vrslev in #68

Full Changelog: 1.3.0...1.3.1

1.3.0

What's Changed

• Add SSL support by @vrslev in #66

Maintenance

• Use faker pytest fixture by @vrslev in #65

Full Changelog: 1.2.0...1.3.0

1.2.0

What's Changed

• Allow passing headers to client.subscribe() by @vrslev in #64
• Use scripts in testing/ as usage examples by @vrslev in #63

Maintenance

• Update development dependencies by @vrslev in #56
• Refactor tests by @vrslev in #57
• Refactor tests (second part) by @vrslev in #58
• Revert tests refactoring by @vrslev in #59
• Allow overriding docker image by @vrslev in #60

Full Changelog: 1.1.0...1.2.0

1.1.0

What's Changed

• Avoid raising RepeatedConnectionLostError when listening frames by @vrslev in #52
• Refactor lifespan and errors by @vrslev in #53 & Fix weird retries by @vrslev in #54

  • Add write_retry_attempts
  • Rename RepeatedConnectionLostError to FailedAllWriteAttemptsError
  • Merge ConnectionConfirmationTimeoutError, UnsupportedProtocolVersionError into FailedAllConnectAttemptsError,
  • and change how FailedAllConnectAttemptsError and FailedAllWriteAttemptsError work (see the PR).

  The most important things are clearer error messages, protocol connection errors (like connection confirmation lost error) included in retries, sane usage of connect_retry_attempts.

• Refactor client modules by @vrslev in #55

  Also: removed internal symbols from stompman/__init__.py, and added dump_frame and FrameParser.

Full Changelog: 1.0.3...1.1.0

1.0.3

What's Changed

• Remove stompman.Client(on_unhandled_message_frame=) by @vrslev in #49
• Fix hanging forever in Client.__aexit__() by @vrslev in #50
• Remove frozen=True and slots=True from exceptions to avoid triggering context manager traceback issue by @vrslev in #51

Full Changelog: 1.0.2...1.0.3

1.0.2

What's Changed

• Return bytes in frame.body instead of bytearray by @vrslev in #48

Full Changelog: 1.0.1...1.0.2

1.0.1

What's Changed

• Add disconnect confirmation timeout by @vrslev in #46
• Force keyword arguments in some cases to avoid ambiguity by @vrslev in #47

Full Changelog: 1.0.0...1.0.1

Changed API, reconnects

What's Changed

• Change transaction API by @vrslev in #42

Before:

async with client.enter_transaction() as transaction:
    await client.send(body=b"hi there!", destination="DLQ", transaction=transaction, headers={"persistent": "true"})

After:

async with client.begin() as transaction:
    await transaction.send(body=b"hi there!", destination="DLQ", headers={"persistent": "true"})

• Change API for subscribing and listening by @vrslev in #43

1. Now, client starts listening to messages the moment you enter the async-with block (async with stompman.Client() as client). Previously it happened only when async for event in Client.listen(): ... was executed. Client.listen() and "events" are removed.
2. To handle ERROR frames and heart-beats, pass handlers to Client().
3. To subscribe to destinations: subscription = await Client.subscribe(destination, handler) (note that it is not a context manager anymore). To unsubscribe: await subscription.unsubscribe().
4. Client will block at end of async-with block if there are any active subscriptions.

See the README.md for more.

Before:

async with asyncio.TaskGroup() as task_group, client.subscribe("DLQ"):
    async for event in client.listen():
        match event:
            case stompman.MessageEvent(body=body):
                task_group.create_task(event.with_auto_ack(handle_message(body), on_suppressed_exception=print))

After:

await client.subscribe("DLQ", handle_message)

(and listening happens in background)

• Implement reconnects on writing and reading frames by @vrslev in #45

Before, stompman.ConnectionLostError was raised each time connection had lost. Now, client will attempt to reconnect, if it fails, stompman.FailedAllConnectAttemptsError will be raised. There's also stompman.RepeatedConnectionLostError for case when connection gets lost when trying to do a single operation like send a frame.
I expect the client to have much more stable behavior after this change.

Maintenance

• Make use of uv run --frozen by @vrslev in #40
• Update README.md by @vrslev in #41
• Bump astral-sh/uv from 0.2.27 to 0.2.31 by @dependabot in #44

New Contributors

• @dependabot made their first contribution in #44

Full Changelog: 0.8.0...1.0.0

0.8.0

What's Changed

• Add property testing for and fix issues regarding ser-de by @vrslev in #27
• Refactor ser-de by @vrslev in #32
• Fixes authorization: Avoid escaping headers in CONNECT frame by @vrslev in #37
• Show received frames in ConnectionConfirmationTimeoutError by @vrslev in #38
• Maybe breaking: Set frozen, kw_only, slots to dataclasses when possible by @vrslev in #39

Maintenance

• Update ruff config by @vrslev in #33
• Use uv docker image by @vrslev in #34
• Unpin uv by @vrslev in #35

Full Changelog: 0.7.1...0.8.0