Stabilising changes for version 3.0.0 (#108)

Author: amunra

.bumpversion.cfg

@@ -0,0 +1,39 @@
+[tool.bumpversion]
+current_version = "3.0.0"
+commit = false
+tag = false
+
+[[tool.bumpversion.files]]
+filename = "pyproject.toml"
+search = "version = \"{current_version}\""
+replace = "version = \"{new_version}\""
+
+[[tool.bumpversion.files]]
+filename = "setup.py"
+search = "version='{current_version}'"
+replace = "version='{new_version}'"
+
+[[tool.bumpversion.files]]
+filename = "README.rst"
+search = "library is {current_version}"
+replace = "library is {new_version}"
+
+[[tool.bumpversion.files]]
+filename = "docs/conf.py"
+search = "version = release = '{current_version}'"
+replace = "version = release = '{new_version}'"
+
+[[tool.bumpversion.files]]
+filename = "src/questdb/__init__.py"
+search = "__version__ = '{current_version}'"
+replace = "__version__ = '{new_version}'"
+
+[[tool.bumpversion.files]]
+filename = "src/questdb/ingress.pyx"
+search = "VERSION = '{current_version}'"
+replace = "VERSION = '{new_version}'"
+
+[[tool.bumpversion.files]]
+filename = ".bumpversion.toml"
+search = "current_version = \"{current_version}\""
+replace = "current_version = \"{new_version}\""

.bumpversion.toml

@@ -5,6 +5,76 @@ Changelog
 
 =========
 
+3.0.0 (2025-07-07)
+------------------
+
+Features
+~~~~~~~~
+
+This is the first major release of the QuestDB Python client library
+which supports n-dimensional arrays of doubles for QuestDB servers 9.0.0 and up.
+
+.. code-block:: python
+
+        import numpy as np
+
+        # Create 2D numpy array
+        array_2d = np.array([
+            [1.1, 2.2, 3.3],
+            [4.4, 5.5, 6.6]], dtype=np.float64)
+
+        sender.row(
+            'table',
+            columns={'array_2d': array_2d},
+            at=timestamp)
+
+The array data is sent over a new protocol version (2) that is auto-negotiated
+when using HTTP(s), or can be specified explicitly via the ``protocol_version=2``
+parameter when using TCP(s).
+
+We recommend using HTTP(s), but here is an TCP example, should you need it::
+
+  tcp::addr=localhost:9009;protocol_version=2;
+
+When using ``protocol_version=2`` (with either TCP(s) or HTTP(s)), the sender
+will now also serialize ``float`` (double-precision) columns as binary.
+You might see a performance uplift if this is a dominant data type in your
+ingestion workload.
+
+When compared to 2.0.4, this release includes all the changes from 3.0.0rc1 and
+additionally:
+
+* Has optimised ingestion performance from C-style contiguous NumPy arrays.
+
+* Warns at most every 10 minutes when burst of reconnections are detected.
+  This is to warn about code patterns that may lead to performance issues, such as
+
+  .. code-block:: python
+
+    # Don't do this! Sender objects should be reused.
+    for row_fields in data:
+        with Sender.from_conf(conf) as sender:
+            sender.row(**row_fields)
+
+  This feature can be disabled in code by setting:
+
+  .. code-block:: python
+
+    import questdb.ingress as qi
+    qi.WARN_HIGH_RECONNECTS = False
+
+* Fixed ILP/TCP connection shutdown on Windows where some rows could be
+  lost when closing the ``Sender``, even if explicitly flushed.
+
+* Added a "Good Practices" section to the "Sending Data over ILP" section of
+  the documentation.
+
+Breaking Changes
+~~~~~~~~~~~~~~~~
+Refer to the release notes for 3.0.0rc1 for the breaking changes introduced
+in this release compared to 2.x.x.
+
+
 3.0.0rc1 (2025-06-02)
 ---------------------
 
@@ -16,6 +86,10 @@ Features
 * Array Data Type Support. Adds native support for NumPy arrays
   (currently only for ``np.float64`` element type and up to 32 dimensions).
 
+.. note::
+    **Server Requirement**: This feature requires QuestDB server version 9.0.0 or higher.
+    Ensure your server is upgraded before ingesting array types, otherwise data ingestion will fail.
+
 .. code-block:: python
 
         import numpy as np

CHANGELOG.rst

@@ -55,7 +55,6 @@ The development requirements are these if you prefer to install them one by one:
     python3 -m pip install wheel
     python3 -m pip install twine
     python3 -m pip install cibuildwheel
-    python3 -m pip install bump2version
 
 
 Building and packaging

DEV_NOTES.rst

@@ -15,7 +15,7 @@ recursive-include src *.py
 recursive-include src *.md
 recursive-include src *.pxi
 recursive-include src *.c
-graft pystr-to-utf8
+graft rpyutils
 graft c-questdb-client
 prune c-questdb-client/src/tests/json_tests.rs
 prune c-questdb-client/.git

MANIFEST.in

@@ -5,7 +5,7 @@ QuestDB Client Library for Python
 This is the official Python client library for `QuestDB <https://questdb.io>`_.
 
 This client library implements QuestDB's variant of the
-`Ingestion Line Protocol <https://questdb.io/docs/reference/api/ilp/overview/>`_
+`InfluxDB Line Protocol <https://questdb.io/docs/reference/api/ilp/overview/>`_
 (ILP) over HTTP and TCP.
 
 ILP provides the fastest way to insert data into QuestDB.
@@ -18,21 +18,12 @@ and full-connection encryption with
 Install
 =======
 
-The latest *stable* version of the library is **2.0.4** (`changelog <https://py-questdb-client.readthedocs.io/en/latest/changelog.html>`_).
+The latest version of the library is **3.0.0** (`changelog <https://py-questdb-client.readthedocs.io/en/latest/changelog.html>`_).
 
 ::
 
     python3 -m pip install -U questdb[dataframe]
 
-
-The latest *pre-release* version of the library is **3.0.0r1** (`changelog <https://py-questdb-client.readthedocs.io/en/latest/changelog.html>`_).
-This release supports NumPy float64 arrays which are transmitted over a new
-protocol version supported by QuestDB 8.4.0 or later.
-
-:: 
-
-    python3 -m pip install --pre -U questdb[dataframe]
-
 Quickstart
 ==========
 
@@ -53,6 +44,7 @@ The most common way to insert data is from a Pandas dataframe.
         'amount': [0.00044, 0.001],
 
         # NumPy float64 arrays are supported from v3.0.0rc1 onwards.
+        # Note that requires QuestDB server >= 9.0.0 for array support
         'ord_book_bids': [
             np.array([2615.54, 2618.63]),
             np.array([39269.98, 39270.00])
@@ -82,6 +74,7 @@ You can also send individual rows. This only requires a more minimal installatio
                 'amount': 0.00044,
 
                 # NumPy float64 arrays are supported from v3.0.0rc1 onwards.
+                # Note that requires QuestDB server >= 9.0.0 for array support
                 'ord_book_bids': np.array([2615.54, 2618.63]),
             },
             at=TimestampNanos.now())

README.rst

@@ -20,21 +20,17 @@ Create a new PR with the new changes in ``CHANGELOG.rst``.
 
 Make a commit and push the changes to a new branch.
 
-You also want to bump the version.
+You also want to bump the version. This process is semi-automated.
 
-This process is automated by the following command:
+* Ensure you have `uv` and `bump-my-version` installed:
+  * `curl -LsSf https://astral.sh/uv/install.sh | sh` : see https://docs.astral.sh/uv/getting-started/installation/
+  * `uv tool install bump-my-version`: see https://github.com/callowayproject/bump-my-version.
 
-.. code-block:: bash
-
-    bump2version --config-file .bumpversion.cfg --no-tag patch
-
-Here use:
-
-* ``patch`` to bump the version to the next patch version, e.g. 1.0.0 -> 1.0.1
-
-* ``minor`` to bump the version to the next minor version, e.g. 1.0.0 -> 1.1.0
+```console
+bump-my-version replace --new-version NEW_VERSION
+```
 
-* ``major`` to bump the version to the next major version, e.g. 1.0.0 -> 2.0.0
+If you're unsure, append `--dry-run` to preview changes.
 
 Now merge the PR with the title "Bump version: V.V.V → W.W.W".
 

RELEASING.rst

@@ -122,6 +122,7 @@ stages:
             displayName: Build wheels
             env:
               CIBW_BUILD: pp*
+              CIBW_ENABLE: pypy pypy-eol
           - task: PublishBuildArtifacts@1
             inputs: {pathtoPublish: 'wheelhouse'}
 

c-questdb-client

@@ -73,6 +73,7 @@ def main(args):
     ensure_timezone()
     pip_install('pip')
     pip_install('setuptools')
+    pip_install('packaging')
     if args.pandas_version is not None and args.pandas_version != '':
         install_old_pandas_and_numpy(args)
     else:
@@ -92,9 +93,7 @@ def main(args):
         import pandas
         import numpy
         import pyarrow
-        if (sys.version_info >= (3, 8) and sys.version_info < (3, 13)):
-            # As of this commit, fastparquet does not have a binary built for 3.13
-            import fastparquet
+        import fastparquet
 
 
 if __name__ == "__main__":