feat: support decimal

ci/run_tests_pipeline.yaml

@@ -36,28 +36,39 @@ stages:
             submodules: true
           - task: UsePythonVersion@0
             inputs:
-              versionSpec: '3.12'
+              versionSpec: "3.12"
           - script: |
               python3 --version
               python3 -m pip install cython
             displayName: "Install cython"
           - script: python3 ci/pip_install_deps.py
             displayName: "Install pandas latest"
             condition: eq(variables.pandasVersion, '')
-          - script:  python3 ci/pip_install_deps.py --pandas-version==$(pandasVersion)
+          - script: python3 ci/pip_install_deps.py --pandas-version==$(pandasVersion)
             displayName: "Install pandas older"
             condition: ne(variables.pandasVersion, '')
           - script: python3 proj.py build
             displayName: "Build"
           - script: |
               git clone --depth 1 https://github.com/questdb/questdb.git
-            displayName: git clone questdb
+            displayName: git clone questdb master
             condition: eq(variables.vsQuestDbMaster, true)
           - task: Maven@3
-            displayName: "Compile QuestDB"
+            displayName: "Compile QuestDB master"
             inputs:
-              mavenPOMFile: 'questdb/pom.xml'
-              jdkVersionOption: '1.17'
+              mavenPOMFile: "questdb/pom.xml"
+              jdkVersionOption: "1.17"
+              options: "-DskipTests -Pbuild-web-console"
+            condition: eq(variables.vsQuestDbMaster, true)
+          - script: |
+              git clone --depth 1 --branch rd_decimal_integration https://github.com/questdb/questdb.git questdb-decimal
+            displayName: git clone questdb decimal branch
+            condition: eq(variables.vsQuestDbMaster, true)
+          - task: Maven@3
+            displayName: "Compile QuestDB decimal branch"
+            inputs:
+              mavenPOMFile: "questdb-decimal/pom.xml"
+              jdkVersionOption: "1.17"
               options: "-DskipTests -Pbuild-web-console"
             condition: eq(variables.vsQuestDbMaster, true)
           - script: python3 proj.py test 1
@@ -68,7 +79,15 @@ stages:
             displayName: "Test vs master"
             env:
               JAVA_HOME: $(JAVA_HOME_17_X64)
-              QDB_REPO_PATH: './questdb'
+              QDB_REPO_PATH: "./questdb"
+            condition: eq(variables.vsQuestDbMaster, true)
+          # Remove before merging decimal support PR
+          - script: python3 proj.py test 1
+            displayName: "Test vs decimal"
+            env:
+              JAVA_HOME: $(JAVA_HOME_17_X64)
+              QDB_REPO_PATH: "./questdb-decimal"
+              TEST_QUESTDB_INTEGRATION_FORCE_MAX_VERSION: "1"
             condition: eq(variables.vsQuestDbMaster, true)
       - job: TestsAgainstVariousNumpyVersion1x
         pool:
@@ -82,7 +101,7 @@ stages:
             submodules: true
           - task: UsePythonVersion@0
             inputs:
-              versionSpec: '3.9'
+              versionSpec: "3.9"
           - script: |
               python3 --version
               python3 -m pip install uv
@@ -106,7 +125,7 @@ stages:
             submodules: true
           - task: UsePythonVersion@0
             inputs:
-              versionSpec: '3.11'
+              versionSpec: "3.11"
           - script: |
               python3 --version
               python3 -m pip install uv

src/questdb/dataframe.md

@@ -93,13 +93,68 @@ We need to extract:
 * 64-bit floats
 * UTF-8 string buffers
 * Nanosecond-precision UTC unix epoch 64-bit signed int timestamps
+* decimals
 
 ```python
 import pandas as pd
 import pyarrow as pa
 import datetime as dt
 ```
 
+### Decimals
+
+Decimals aren't natively supported by pandas nor numpy, they use the `decimal.Decimal` objects.
+
+#### Pandas
+
+Decimals stored as Python objects in an 'object' dtype column.
+
+```python
+>>> df = pd.DataFrame({ 'decimals': [Decimal('123.456')] })
+>>> df.dtypes
+decimals    object
+dtype: object
+```
+
+#### Numpy
+
+Similarly, numpy stores decimals as Python objects.
+
+```python
+>>> arr = numpy.array([Decimal('123.456')])
+>>> arr
+array([Decimal('123.456')], dtype=object)
+```
+
+#### PyArrow
+
+PyArrow provides native decimal support with configurable precision and scale.
+The data is stored in a fixed-width binary format.
+
+```python
+import pyarrow as pa
+from decimal import Decimal
+
+# Create decimal array: decimal128(precision, scale)
+# precision = total digits, scale = digits after decimal point
+decimal_array = pa.array(
+    [Decimal('123.456'), Decimal('789.012'), Decimal('-456.789'), None],
+    type=pa.decimal128(10, 3)  # 10 total digits, 3 after decimal
+)
+
+# Use in DataFrame with ArrowDtype
+df = pd.DataFrame({
+    'prices': pd.array(
+        [Decimal('123.45'), Decimal('678.90'), None],
+        dtype=pd.ArrowDtype(pa.decimal128(10, 2))
+    )
+})
+```
+
+Notes:
+- 4 datatypes: `decimal32`, `decimal64`, `decimal128` and `decimal256`
+- Nulls are supported via Arrow's validity bitmap
+
 ### Booleans
 
 ```python