sqlmath 2026.7.30 → 2026.9.29

package/CHANGELOG.md

@@ -7,6 +7,22 @@
 - sqlite - Add similar error-handling from builtin-sql-function PERCENTILE() into custom-sql-function QUANTILE().
 - none
 
+# v2026.9.29
+- sqlmath - Update sql-function WIN_COINFLIP2() to calculate 'biasba50' - balanced-accuracy-minus-50.
+- jslint-ecma - Add Exponentiation-assignment-operator '**=' support.
+- jslint-cli - Add cli-option jslint_autofix=filename to autofix whitespace-warnings, and add function jslint_phase6_autofix().
+- jslint-ecma - Add ES2018-feature Asynchronous Iteration - for await...of.
+- jslint-cli - Skip dotfiles when linting a directory.
+
+# v2026.8.31
+- jslint-ci - Update ci-shell-function shGitLsTree() to include sha256 hash of files.
+- sqlmath - Cleanup c-function winSinefitSnr().
+- sqlmath - Update sql-function WIN_SINEFIT2() argument modeSnr with bitmask, allowing selective sine-fit of each column, to avoid wasting cpu on columns that don't need it.
+- lgbm - Upgrade to lightgbm-v4.7.0.
+- sqlite - Upgrade to sqlite-v3.51.3.
+- sqlmath - Remove obsolete sql-functions MEDIAN2(), QUANTILE(), superseded by sqlite-builtins.
+- sqlmath - Fix c-header conflict between dlfcn.h and Python.h.
+
 # v2026.7.30
 - jslint-ecma - Add ES2015-feature for..of.
 - jslint - Change scope from scope_function to scope_block:

package/README.md

@@ -9,7 +9,7 @@
 
 # Status
 
-| Branch | [master<br>(v2026.7.30)](https://github.com/sqlmath/sqlmath/tree/master) | [beta<br>(Web Demo)](https://github.com/sqlmath/sqlmath/tree/beta) | [alpha<br>(Development)](https://github.com/sqlmath/sqlmath/tree/alpha) |
+| Branch | [master<br>(v2026.9.29)](https://github.com/sqlmath/sqlmath/tree/master) | [beta<br>(Web Demo)](https://github.com/sqlmath/sqlmath/tree/beta) | [alpha<br>(Development)](https://github.com/sqlmath/sqlmath/tree/alpha) |
 |--:|:--:|:--:|:--:|
 | CI | [![ci](https://github.com/sqlmath/sqlmath/actions/workflows/ci.yml/badge.svg?branch=master)](https://github.com/sqlmath/sqlmath/actions?query=branch%3Amaster) | [![ci](https://github.com/sqlmath/sqlmath/actions/workflows/ci.yml/badge.svg?branch=beta)](https://github.com/sqlmath/sqlmath/actions?query=branch%3Abeta) | [![ci](https://github.com/sqlmath/sqlmath/actions/workflows/ci.yml/badge.svg?branch=alpha)](https://github.com/sqlmath/sqlmath/actions?query=branch%3Aalpha) |
 | Coverage | [![coverage](https://sqlmath.github.io/sqlmath/branch-master/.artifact/coverage/coverage_badge.svg)](https://sqlmath.github.io/sqlmath/branch-master/.artifact/coverage/index.html) | [![coverage](https://sqlmath.github.io/sqlmath/branch-beta/.artifact/coverage/coverage_badge.svg)](https://sqlmath.github.io/sqlmath/branch-beta/.artifact/coverage/index.html) | [![coverage](https://sqlmath.github.io/sqlmath/branch-alpha/.artifact/coverage/coverage_badge.svg)](https://sqlmath.github.io/sqlmath/branch-alpha/.artifact/coverage/index.html) |
@@ -18,6 +18,7 @@
 
 
 <br><br>
+
 # Table of Contents
 
 1. [Web Demo](#web-demo)
@@ -78,6 +79,7 @@
 
 
 <br><br>
+
 # Web Demo
 - https://sqlmath.github.io/sqlmath/index.html
 
@@ -85,6 +87,7 @@
 
 
 <br><br>
+
 # Why sqlmath?
 
 SQLite is everywhere — it's the most deployed database in the world. But it lacks the statistical and ML functions needed for data-science. sqlmath fixes that.
@@ -97,10 +100,12 @@ SQLite is everywhere — it's the most deployed database in the world. But it la
 
 
 <br><br>
+
 # Quickstart
 
 
 <br><br>
+
 ### Python
 
 ```shell
@@ -135,6 +140,7 @@ db_close(db)
 
 
 <br><br>
+
 ### JavaScript (Node.js)
 
 ```shell
@@ -165,16 +171,19 @@ await dbCloseAsync(db);
 
 
 <br><br>
+
 ### Browser (WebAssembly)
 
 Try it live: **[sqlmath.github.io/sqlmath](https://sqlmath.github.io/sqlmath/index.html)**
 
 
 <br><br>
+
 # Built-in SQL Functions
 
 
 <br><br>
+
 ### Math
 
 | Function | Description |
@@ -190,6 +199,7 @@ Try it live: **[sqlmath.github.io/sqlmath](https://sqlmath.github.io/sqlmath/ind
 
 
 <br><br>
+
 ### Statistics (Aggregate)
 
 | Function | Description |
@@ -201,6 +211,7 @@ Try it live: **[sqlmath.github.io/sqlmath](https://sqlmath.github.io/sqlmath/ind
 
 
 <br><br>
+
 ### Statistics (Scalar)
 
 | Function | Description |
@@ -210,6 +221,7 @@ Try it live: **[sqlmath.github.io/sqlmath](https://sqlmath.github.io/sqlmath/ind
 
 
 <br><br>
+
 ### Arrays
 
 | Function | Description |
@@ -221,12 +233,14 @@ Try it live: **[sqlmath.github.io/sqlmath](https://sqlmath.github.io/sqlmath/ind
 
 
 <br><br>
+
 ### Date/Time
 
 sqlmath extends SQLite's date functions with integer-format conversions (YYYYMMDDHHMMSS).
 
 
 <br><br>
+
 ### Time Series / Signal Processing
 
 | Function | Description |
@@ -236,6 +250,7 @@ sqlmath extends SQLite's date functions with integer-format conversions (YYYYMMD
 
 
 <br><br>
+
 ### Compression
 
 | Function | Description |
@@ -247,6 +262,7 @@ sqlmath extends SQLite's date functions with integer-format conversions (YYYYMMD
 
 
 <br><br>
+
 ### Type Casting
 
 | Function | Description |
@@ -258,6 +274,7 @@ sqlmath extends SQLite's date functions with integer-format conversions (YYYYMMD
 
 
 <br><br>
+
 ### Cryptography
 
 | Function | Description |
@@ -271,6 +288,7 @@ SELECT HEX(sha256('hello')) AS hash;
 
 
 <br><br>
+
 # Machine Learning with LightGBM
 
 sqlmath embeds [LightGBM](https://lightgbm.readthedocs.io/) for gradient boosting directly in SQL queries. Train models on your data without leaving SQL — no data shuffle to Python needed.
@@ -279,6 +297,7 @@ sqlmath embeds [LightGBM](https://lightgbm.readthedocs.io/) for gradient boostin
 
 
 <br><br>
+
 ### Training from a Table
 
 `LGBM_TRAINFROMTABLE` is an **aggregate function** — it consumes rows like `SUM()` or `AVG()`, but outputs a trained model BLOB.
@@ -311,6 +330,7 @@ FROM training_data;
 
 
 <br><br>
+
 ### Prediction
 
 ```sql
@@ -330,6 +350,7 @@ FROM test_data;
 
 
 <br><br>
+
 ### Real-World Example: Credit Card Fraud Detection
 
 From the [Kaggle notebook](https://www.kaggle.com/code/kaizhu256/sql-is-all-you-need) — training on 284,807 transactions with 0.17% fraud rate:
@@ -378,6 +399,7 @@ db_exec(db=db, sql="""
 
 
 <br><br>
+
 ### LightGBM Functions
 
 | Function | Type | Description |
@@ -398,14 +420,16 @@ Full notebook with fraud detection, intraday trading signals, and model persiste
 **Kaggle Environment:**
 - Linux x64, Python 3.12, Node.js 24
 - Datasets: [Credit Card Fraud](https://www.kaggle.com/datasets/mlg-ulb/creditcardfraud) (284K transactions), SPY intraday OHLCV
-- sqlmath installs via `pip install sqlmath==2026.7.30`
+- sqlmath installs via `pip install sqlmath==2026.9.29`
 
 
 <br><br>
+
 # Use Cases
 
 
 <br><br>
+
 ### Financial Data Analysis
 
 ```python
@@ -434,6 +458,7 @@ db_close(db)
 
 
 <br><br>
+
 ### Embedded ML Pipelines
 
 Train and deploy models without data movement:
@@ -460,6 +485,7 @@ db_file_save(db=db, filename="scored_data.sqlite")
 
 
 <br><br>
+
 ### Data Compression & Hashing
 
 ```sql
@@ -476,6 +502,7 @@ HAVING COUNT(*) > 1;
 
 
 <br><br>
+
 # Why sqlmath vs Alternatives?
 
 | Feature | sqlmath | pandas | DuckDB | sqlite3 |
@@ -483,14 +510,14 @@ HAVING COUNT(*) > 1;
 | In-database ML | ✅ LightGBM | ❌ | ❌ | ❌ |
 | Zero data shuffle | ✅ | ❌ | ✅ | ✅ |
 | Statistical functions | ✅ 20+ | ✅ | ✅ | ❌ |
-| Browser support | ✅ WASM | ❌ | ❌ | ❌ |
+| Browser support | ✅ WASM + LightGBM | ❌ | ✅ DuckDB-Wasm | ✅ official WASM |
 | Single file deployment | ✅ | ❌ | ✅ | ✅ |
 | Memory efficiency | ✅ Streaming | ❌ In-memory | ✅ | ✅ |
 
 **Choose sqlmath when:**
 - You need ML training/inference inside the database
 - Your data is already in SQLite
-- You want browser-based analytics (WebAssembly)
+- You want ML training/inference in the browser (WebAssembly)
 - You need statistical functions beyond basic SQL
 
 **Choose pandas when:**
@@ -505,6 +532,7 @@ HAVING COUNT(*) > 1;
 
 
 <br><br>
+
 # Python API Reference
 
 ```python
@@ -522,6 +550,7 @@ from sqlmath import (
 
 
 <br><br>
+
 ### db_exec()
 
 ```python
@@ -537,6 +566,7 @@ result = db_exec(
 
 
 <br><br>
+
 # Node.js API Reference
 - https://sqlmath.github.io/sqlmath/apidoc.html
 
@@ -544,6 +574,7 @@ result = db_exec(
 
 
 <br><br>
+
 # Platform Notes
 
 The JavaScript (Node.js) binding is more mature than Python. Key differences:
@@ -559,10 +590,12 @@ The JavaScript (Node.js) binding is more mature than Python. Key differences:
 
 
 <br><br>
+
 # Building from Source
 
 
 <br><br>
+
 ### Prerequisites
 
 - Node.js 24+
@@ -571,6 +604,7 @@ The JavaScript (Node.js) binding is more mature than Python. Key differences:
 
 
 <br><br>
+
 ### Build
 
 ```shell
@@ -589,6 +623,7 @@ sh jslint_ci.sh shCiBuildWasm
 
 
 <br><br>
+
 ### Run Tests
 
 ```shell
@@ -604,6 +639,7 @@ npm run test --fast
 
 
 <br><br>
+
 ### Serve Demo Locally
 
 ```shell
@@ -613,11 +649,13 @@ PORT=8080 sh jslint_ci.sh shHttpFileServer
 
 
 <br><br>
+
 # Package Listing
 ![screenshot_package_listing.svg](https://sqlmath.github.io/sqlmath/branch-beta/.artifact/screenshot_package_listing.svg)
 
 
 <br><br>
+
 # Changelog
 - [Full CHANGELOG.md](CHANGELOG.md)
 
@@ -625,6 +663,7 @@ PORT=8080 sh jslint_ci.sh shHttpFileServer
 
 
 <br><br>
+
 # License
 - [sqlite](https://github.com/sqlite/sqlite) is under [public domain](https://www.sqlite.org/copyright.html).
 - [jslint](https://github.com/jslint-org/jslint) is under [Unlicense License](https://github.com/jslint-org/jslint/blob/master/LICENSE).
@@ -635,34 +674,37 @@ PORT=8080 sh jslint_ci.sh shHttpFileServer
 
 
 <br><br>
+
 # Devops Instruction
 
 
 <br><br>
+
 ### python pypi publish
 ```shell
 python -m build
 #
-twine upload --repository testpypi dist/sqlmath-2026.7.30*
-py -m pip install --index-url https://test.pypi.org/simple/ sqlmath==2026.7.30
+twine upload --repository testpypi dist/sqlmath-2026.9.29*
+py -m pip install --index-url https://test.pypi.org/simple/ sqlmath==2026.9.29
 #
-twine upload dist/sqlmath-2026.7.30*
-pip install sqlmath==2026.7.30
+twine upload dist/sqlmath-2026.9.29*
+pip install sqlmath==2026.9.29
 ```
 
 
 <br><br>
+
 ### sqlite upgrade
 - goto https://www.sqlite.org/changes.html
 ```shell
     (set -e
     #
     # lgbm
-    sh jslint_ci.sh shRollupUpgrade "v4.5.0" "v4.6.0" ".ci.sh sqlmath_base.h"
+    sh jslint_ci.sh shRollupUpgrade "v4.6.0" "v4.7.0" ".ci.sh sqlmath_base.h"
     #
     # sqlite
-    sh jslint_ci.sh shRollupUpgrade "3.50.3" "3.50.4" ".ci.sh sqlmath_external_sqlite.c"
-    sh jslint_ci.sh shRollupUpgrade "3500300" "3500400" ".ci.sh sqlmath_external_sqlite.c"
+    sh jslint_ci.sh shRollupUpgrade "3.50.4" "3.51.3" ".ci.sh sqlmath_external_sqlite.c"
+    sh jslint_ci.sh shRollupUpgrade "3500400" "3510300" ".ci.sh sqlmath_external_sqlite.c"
     #
     # shSqlmathUpdate
     read -p "Press Enter to shSqlmathUpdate:"