koffi 1.3.5 → 1.3.8

package/doc/dist/html/benchmarks.html

@@ -5,7 +5,7 @@
     <meta name="color-scheme" content="light dark"><meta name="generator" content="Docutils 0.18.1: http://docutils.sourceforge.net/" />
 <link rel="index" title="Index" href="genindex" /><link rel="search" title="Search" href="search" /><link rel="next" title="Contributing" href="contribute" /><link rel="prev" title="Memory usage" href="memory" />
 
-    <meta name="generator" content="sphinx-5.0.1, furo 2022.06.04.1"/>
+    <meta name="generator" content="sphinx-5.0.2, furo 2022.06.21"/>
         <title>Benchmarks - Koffi</title>
       <link rel="stylesheet" type="text/css" href="_static/pygments.css" />
     <link rel="stylesheet" type="text/css" href="_static/styles/furo.css?digest=40978830699223671f4072448e654b5958f38b89" />
@@ -162,7 +162,7 @@
 </form>
 <div id="searchbox"></div><div class="sidebar-scroll"><div class="sidebar-tree">
   <ul class="current">
-<li class="toctree-l1"><a class="reference internal" href="platforms">Supported platforms</a></li>
+<li class="toctree-l1"><a class="reference internal" href="platforms">Requirements</a></li>
 <li class="toctree-l1"><a class="reference internal" href="start">Quick start</a></li>
 <li class="toctree-l1"><a class="reference internal" href="types">Data types</a></li>
 <li class="toctree-l1"><a class="reference internal" href="functions">Function calls</a></li>
@@ -205,6 +205,8 @@
         <article role="main">
           <section id="benchmarks">
 <h1>Benchmarks<a class="headerlink" href="#benchmarks" title="Permalink to this heading">#</a></h1>
+<section id="overview">
+<h2>Overview<a class="headerlink" href="#overview" title="Permalink to this heading">#</a></h2>
 <p>Here is a quick overview of the execution time of Koffi calls on three benchmarks, where it is compared to a theoretical ideal FFI implementation (approximated with pre-compiled static N-API glue code):</p>
 <ul class="simple">
 <li><p>The first benchmark is based on <code class="docutils literal notranslate"><span class="pre">rand()</span></code> calls</p></li>
@@ -213,272 +215,261 @@
 </ul>
 <table style="margin: 0 auto;">
     <tr>
-        <td><a href="_static/perf_linux_20220623_2.png" target="_blank"><img src="_static/perf_linux_20220623_2.png" alt="Linux performance" style="width: 350px;"/></a></td>
-        <td><a href="_static/perf_windows_20220623_2.png" target="_blank"><img src="_static/perf_windows_20220623_2.png" alt="Windows performance" style="width: 350px;"/></a></td>
+        <td><a href="_static/perf_linux_20220628.png" target="_blank"><img src="_static/perf_linux_20220628.png" alt="Linux performance" style="width: 350px;"/></a></td>
+        <td><a href="_static/perf_windows_20220628.png" target="_blank"><img src="_static/perf_windows_20220628.png" alt="Windows performance" style="width: 350px;"/></a></td>
     </tr>
 </table>
 <p>These results are detailed and explained below, and compared to node-ffi/node-ffi-napi.</p>
+</section>
+<section id="linux-x86-64">
+<h2>Linux x86_64<a class="headerlink" href="#linux-x86-64" title="Permalink to this heading">#</a></h2>
+<p>The results presented below were measured on my x86_64 Linux machine (Intel® Core™ i5-4460).</p>
 <section id="rand-results">
-<h2>rand results<a class="headerlink" href="#rand-results" title="Permalink to this heading">#</a></h2>
+<h3>rand results<a class="headerlink" href="#rand-results" title="Permalink to this heading">#</a></h3>
 <p>This test is based around repeated calls to a simple standard C function atoi, and has three implementations:</p>
 <ul class="simple">
 <li><p>the first one is the reference, it calls atoi through an N-API module, and is close to the theoretical limit of a perfect (no overhead) Node.js &gt; C FFI implementation (pre-compiled static glue code)</p></li>
 <li><p>the second one calls atoi through Koffi</p></li>
 <li><p>the third one uses the official Node.js FFI implementation, node-ffi-napi</p></li>
 </ul>
-<p>Because rand is a pretty small function, the FFI overhead is clearly visible.</p>
-<section id="linux-x86-64">
-<h3>Linux x86_64<a class="headerlink" href="#linux-x86-64" title="Permalink to this heading">#</a></h3>
-<p>The results below were measured on my x86_64 Linux machine (AMD® Ryzen™ 7 4700U):</p>
 <div class="table-wrapper colwidths-auto docutils container">
 <table class="docutils align-default">
 <thead>
 <tr class="row-odd"><th class="head"><p>Benchmark</p></th>
-<th class="head"><p>Iterations</p></th>
-<th class="head"><p>Total time</p></th>
+<th class="head"><p>Iteration time</p></th>
 <th class="head"><p>Relative performance</p></th>
 <th class="head"><p>Overhead</p></th>
 </tr>
 </thead>
 <tbody>
 <tr class="row-even"><td><p>rand_napi</p></td>
-<td><p>20000000</p></td>
-<td><p>1.44s</p></td>
-<td><p>(baseline)</p></td>
-<td><p>(baseline)</p></td>
+<td><p>644 ns</p></td>
+<td><p>x1.00</p></td>
+<td><p>(ref)</p></td>
 </tr>
 <tr class="row-odd"><td><p>rand_koffi</p></td>
-<td><p>20000000</p></td>
-<td><p>2.60s</p></td>
-<td><p>x0.55</p></td>
-<td><p>+81%</p></td>
+<td><p>950 ns</p></td>
+<td><p>x0.68</p></td>
+<td><p>+48%</p></td>
 </tr>
 <tr class="row-even"><td><p>rand_node_ffi</p></td>
-<td><p>20000000</p></td>
-<td><p>107.58s</p></td>
-<td><p>x0.01</p></td>
-<td><p>+7400%</p></td>
+<td><p>30350 ns</p></td>
+<td><p>x0.02</p></td>
+<td><p>+4613%</p></td>
 </tr>
 </tbody>
 </table>
 </div>
+<p>Because rand is a pretty small function, the FFI overhead is clearly visible.</p>
 </section>
-<section id="windows-x86-64">
-<h3>Windows x86_64<a class="headerlink" href="#windows-x86-64" title="Permalink to this heading">#</a></h3>
-<p>The results below were measured on my x86_64 Windows machine (Intel® Core™ i5-4460):</p>
+<section id="atoi-results">
+<h3>atoi results<a class="headerlink" href="#atoi-results" title="Permalink to this heading">#</a></h3>
+<p>This test is similar to the rand one, but it is based on atoi, which takes a string parameter. Javascript (V8) to C string conversion is relatively slow and heavy.</p>
 <div class="table-wrapper colwidths-auto docutils container">
 <table class="docutils align-default">
 <thead>
 <tr class="row-odd"><th class="head"><p>Benchmark</p></th>
-<th class="head"><p>Iterations</p></th>
-<th class="head"><p>Total time</p></th>
+<th class="head"><p>Iteration time</p></th>
 <th class="head"><p>Relative performance</p></th>
 <th class="head"><p>Overhead</p></th>
 </tr>
 </thead>
 <tbody>
-<tr class="row-even"><td><p>rand_napi</p></td>
-<td><p>20000000</p></td>
-<td><p>2.10s</p></td>
-<td><p>(baseline)</p></td>
-<td><p>(baseline)</p></td>
+<tr class="row-even"><td><p>atoi_napi</p></td>
+<td><p>1104 ns</p></td>
+<td><p>x1.00</p></td>
+<td><p>(ref)</p></td>
 </tr>
-<tr class="row-odd"><td><p>rand_koffi</p></td>
-<td><p>20000000</p></td>
-<td><p>3.87s</p></td>
-<td><p>x0.54</p></td>
-<td><p>+84%</p></td>
+<tr class="row-odd"><td><p>atoi_koffi</p></td>
+<td><p>1778 ns</p></td>
+<td><p>x0.62</p></td>
+<td><p>+61%</p></td>
 </tr>
-<tr class="row-even"><td><p>rand_node_ffi</p></td>
-<td><p>20000000</p></td>
-<td><p>87.84s</p></td>
-<td><p>x0.02</p></td>
-<td><p>+4100%</p></td>
+<tr class="row-even"><td><p>atoi_node_ffi</p></td>
+<td><p>125300 ns</p></td>
+<td><p>x0.009</p></td>
+<td><p>+11250%</p></td>
 </tr>
 </tbody>
 </table>
 </div>
+<p>Because atoi is a pretty small function, the FFI overhead is clearly visible.</p>
 </section>
-</section>
-<section id="atoi-results">
-<h2>atoi results<a class="headerlink" href="#atoi-results" title="Permalink to this heading">#</a></h2>
-<p>This test is similar to the rand one, but it is based on atoi, which takes a string parameter. Javascript (V8) to C string conversion is relatively slow and heavy.</p>
-<p>Because rand is a pretty small function, the FFI overhead is clearly visible.</p>
-<section id="id1">
-<h3>Linux x86_64<a class="headerlink" href="#id1" title="Permalink to this heading">#</a></h3>
-<p>The results below were measured on my x86_64 Linux machine (AMD® Ryzen™ 7 4700U):</p>
+<section id="raylib-results">
+<h3>Raylib results<a class="headerlink" href="#raylib-results" title="Permalink to this heading">#</a></h3>
+<p>This benchmark uses the CPU-based image drawing functions in Raylib. The calls are much heavier than in the atoi benchmark, thus the FFI overhead is reduced. In this implementation, Koffi is compared to:</p>
+<ul class="simple">
+<li><p>Baseline: Full C++ version of the code (no JS)</p></li>
+<li><p><a class="reference external" href="https://github.com/RobLoach/node-raylib">node-raylib</a>: This is a native wrapper implemented with N-API</p></li>
+</ul>
 <div class="table-wrapper colwidths-auto docutils container">
 <table class="docutils align-default">
 <thead>
 <tr class="row-odd"><th class="head"><p>Benchmark</p></th>
-<th class="head"><p>Iterations</p></th>
-<th class="head"><p>Total time</p></th>
+<th class="head"><p>Iteration time</p></th>
 <th class="head"><p>Relative performance</p></th>
 <th class="head"><p>Overhead</p></th>
 </tr>
 </thead>
 <tbody>
-<tr class="row-even"><td><p>atoi_napi</p></td>
-<td><p>20000000</p></td>
-<td><p>2.97s</p></td>
-<td><p>(baseline)</p></td>
-<td><p>(baseline)</p></td>
+<tr class="row-even"><td><p>raylib_cc</p></td>
+<td><p>215.7 µs</p></td>
+<td><p>x1.20</p></td>
+<td><p>-17%</p></td>
 </tr>
-<tr class="row-odd"><td><p>atoi_koffi</p></td>
-<td><p>20000000</p></td>
-<td><p>5.07s</p></td>
-<td><p>x0.58</p></td>
-<td><p>+71%</p></td>
+<tr class="row-odd"><td><p>raylib_node_raylib</p></td>
+<td><p>258.9 µs</p></td>
+<td><p>x1.00</p></td>
+<td><p>(ref)</p></td>
 </tr>
-<tr class="row-even"><td><p>atoi_node_ffi</p></td>
-<td><p>20000000</p></td>
-<td><p>693.16s</p></td>
-<td><p>x0.005</p></td>
-<td><p>+23000%</p></td>
+<tr class="row-even"><td><p>raylib_koffi</p></td>
+<td><p>311.6 µs</p></td>
+<td><p>x0.83</p></td>
+<td><p>+20%</p></td>
+</tr>
+<tr class="row-odd"><td><p>raylib_node_ffi</p></td>
+<td><p>928.4 µs</p></td>
+<td><p>x0.28</p></td>
+<td><p>+259%</p></td>
 </tr>
 </tbody>
 </table>
 </div>
 </section>
-<section id="id2">
-<h3>Windows x86_64<a class="headerlink" href="#id2" title="Permalink to this heading">#</a></h3>
-<p>The results below were measured on my x86_64 Windows machine (Intel® Core™ i5-4460):</p>
+</section>
+<section id="windows-x86-64">
+<h2>Windows x86_64<a class="headerlink" href="#windows-x86-64" title="Permalink to this heading">#</a></h2>
+<p>The results presented below were measured on my x86_64 Windows machine (Intel® Core™ i5-4460).</p>
+<section id="id1">
+<h3>rand results<a class="headerlink" href="#id1" title="Permalink to this heading">#</a></h3>
+<p>This test is based around repeated calls to a simple standard C function atoi, and has three implementations:</p>
+<ul class="simple">
+<li><p>the first one is the reference, it calls atoi through an N-API module, and is close to the theoretical limit of a perfect (no overhead) Node.js &gt; C FFI implementation (pre-compiled static glue code)</p></li>
+<li><p>the second one calls atoi through Koffi</p></li>
+<li><p>the third one uses the official Node.js FFI implementation, node-ffi-napi</p></li>
+</ul>
 <div class="table-wrapper colwidths-auto docutils container">
 <table class="docutils align-default">
 <thead>
 <tr class="row-odd"><th class="head"><p>Benchmark</p></th>
-<th class="head"><p>Iterations</p></th>
-<th class="head"><p>Total time</p></th>
+<th class="head"><p>Iteration time</p></th>
 <th class="head"><p>Relative performance</p></th>
 <th class="head"><p>Overhead</p></th>
 </tr>
 </thead>
 <tbody>
-<tr class="row-even"><td><p>atoi_napi</p></td>
-<td><p>20000000</p></td>
-<td><p>2.97s</p></td>
-<td><p>(baseline)</p></td>
-<td><p>(baseline)</p></td>
+<tr class="row-even"><td><p>rand_napi</p></td>
+<td><p>965 ns</p></td>
+<td><p>x1.00</p></td>
+<td><p>(ref)</p></td>
 </tr>
-<tr class="row-odd"><td><p>atoi_koffi</p></td>
-<td><p>20000000</p></td>
-<td><p>5.91s</p></td>
-<td><p>x0.50</p></td>
-<td><p>+99%</p></td>
+<tr class="row-odd"><td><p>rand_koffi</p></td>
+<td><p>1248 ns</p></td>
+<td><p>x0.77</p></td>
+<td><p>+29%</p></td>
 </tr>
-<tr class="row-even"><td><p>atoi_node_ffi</p></td>
-<td><p>20000000</p></td>
-<td><p>479.34s</p></td>
-<td><p>x0.006</p></td>
-<td><p>+16000%</p></td>
+<tr class="row-even"><td><p>rand_node_ffi</p></td>
+<td><p>41500 ns</p></td>
+<td><p>x0.02</p></td>
+<td><p>+4203%</p></td>
 </tr>
 </tbody>
 </table>
 </div>
+<p>Because rand is a pretty small function, the FFI overhead is clearly visible.</p>
 </section>
-</section>
-<section id="raylib-results">
-<h2>Raylib results<a class="headerlink" href="#raylib-results" title="Permalink to this heading">#</a></h2>
-<p>This benchmark uses the CPU-based image drawing functions in Raylib. The calls are much heavier than in the atoi benchmark, thus the FFI overhead is reduced. In this implementation, Koffi is compared to:</p>
-<ul class="simple">
-<li><p>Baseline: Full C++ version of the code (no JS)</p></li>
-<li><p><a class="reference external" href="https://github.com/RobLoach/node-raylib">node-raylib</a>: This is a native wrapper implemented with N-API</p></li>
-</ul>
-<section id="id3">
-<h3>Linux x86_64<a class="headerlink" href="#id3" title="Permalink to this heading">#</a></h3>
-<p>The results below were measured on my x86_64 Linux machine (AMD® Ryzen™ 7 4700U):</p>
+<section id="id2">
+<h3>atoi results<a class="headerlink" href="#id2" title="Permalink to this heading">#</a></h3>
+<p>This test is similar to the rand one, but it is based on atoi, which takes a string parameter. Javascript (V8) to C string conversion is relatively slow and heavy.</p>
+<p>The results below were measured on my x86_64 Windows machine (Intel® Core™ i5-4460):</p>
 <div class="table-wrapper colwidths-auto docutils container">
 <table class="docutils align-default">
 <thead>
 <tr class="row-odd"><th class="head"><p>Benchmark</p></th>
-<th class="head"><p>Iterations</p></th>
-<th class="head"><p>Total time</p></th>
+<th class="head"><p>Iteration time</p></th>
 <th class="head"><p>Relative performance</p></th>
 <th class="head"><p>Overhead</p></th>
 </tr>
 </thead>
 <tbody>
-<tr class="row-even"><td><p>raylib_cc</p></td>
-<td><p>100</p></td>
-<td><p>9.31s</p></td>
-<td><p>x1.17</p></td>
-<td><p>-15%</p></td>
-</tr>
-<tr class="row-odd"><td><p>raylib_node_raylib</p></td>
-<td><p>100</p></td>
-<td><p>10.90s</p></td>
-<td><p>(baseline)</p></td>
-<td><p>(baseline)</p></td>
+<tr class="row-even"><td><p>atoi_napi</p></td>
+<td><p>1393 ns</p></td>
+<td><p>x1.00</p></td>
+<td><p>(ref)</p></td>
 </tr>
-<tr class="row-even"><td><p>raylib_koffi</p></td>
-<td><p>100</p></td>
-<td><p>12.86s</p></td>
-<td><p>x0.84</p></td>
-<td><p>+18%</p></td>
+<tr class="row-odd"><td><p>atoi_koffi</p></td>
+<td><p>2246 ns</p></td>
+<td><p>x0.62</p></td>
+<td><p>+61%</p></td>
 </tr>
-<tr class="row-odd"><td><p>raylib_node_ffi</p></td>
-<td><p>100</p></td>
-<td><p>35.76s</p></td>
-<td><p>x0.30</p></td>
-<td><p>+228%</p></td>
+<tr class="row-even"><td><p>atoi_node_ffi</p></td>
+<td><p>157550 ns</p></td>
+<td><p>x0.009</p></td>
+<td><p>+11210%</p></td>
 </tr>
 </tbody>
 </table>
 </div>
+<p>Because atoi is a pretty small function, the FFI overhead is clearly visible.</p>
 </section>
-<section id="id4">
-<h3>Windows x86_64<a class="headerlink" href="#id4" title="Permalink to this heading">#</a></h3>
-<p>The results below were measured on my x86_64 Windows machine (Intel® Core™ i5-4460):</p>
+<section id="id3">
+<h3>Raylib results<a class="headerlink" href="#id3" title="Permalink to this heading">#</a></h3>
+<p>This benchmark uses the CPU-based image drawing functions in Raylib. The calls are much heavier than in the atoi benchmark, thus the FFI overhead is reduced. In this implementation, Koffi is compared to:</p>
+<ul class="simple">
+<li><p><a class="reference external" href="https://github.com/RobLoach/node-raylib">node-raylib</a> (baseline): This is a native wrapper implemented with N-API</p></li>
+<li><p>raylib_cc: C++ implementation of the benchmark, without any Javascript</p></li>
+</ul>
 <div class="table-wrapper colwidths-auto docutils container">
 <table class="docutils align-default">
 <thead>
 <tr class="row-odd"><th class="head"><p>Benchmark</p></th>
-<th class="head"><p>Iterations</p></th>
-<th class="head"><p>Total time</p></th>
+<th class="head"><p>Iteration time</p></th>
 <th class="head"><p>Relative performance</p></th>
 <th class="head"><p>Overhead</p></th>
 </tr>
 </thead>
 <tbody>
 <tr class="row-even"><td><p>raylib_cc</p></td>
-<td><p>100</p></td>
-<td><p>10.67s</p></td>
-<td><p>x1.17</p></td>
-<td><p>-12%</p></td>
+<td><p>211.8 µs</p></td>
+<td><p>x1.25</p></td>
+<td><p>-20%</p></td>
 </tr>
 <tr class="row-odd"><td><p>raylib_node_raylib</p></td>
-<td><p>100</p></td>
-<td><p>12.05s</p></td>
-<td><p>(baseline)</p></td>
-<td><p>(baseline)</p></td>
+<td><p>264.4 µs</p></td>
+<td><p>x1.00</p></td>
+<td><p>(ref)</p></td>
 </tr>
 <tr class="row-even"><td><p>raylib_koffi</p></td>
-<td><p>100</p></td>
-<td><p>14.84s</p></td>
-<td><p>x0.81</p></td>
-<td><p>+23%</p></td>
+<td><p>318.9 µs</p></td>
+<td><p>x0.83</p></td>
+<td><p>+21%</p></td>
 </tr>
 <tr class="row-odd"><td><p>raylib_node_ffi</p></td>
-<td><p>100</p></td>
-<td><p>44.63s</p></td>
-<td><p>x0.27</p></td>
-<td><p>+270%</p></td>
+<td><p>1146.2 µs</p></td>
+<td><p>x0.23</p></td>
+<td><p>+334%</p></td>
 </tr>
 </tbody>
 </table>
 </div>
+<p>Please note that in order to get fair numbers for raylib_node_raylib, it was recompiled with clang-cl before running the benchmark with the following commands:</p>
+<div class="highlight-batch notranslate"><div class="highlight"><pre><span></span><span class="linenos">1</span><span class="k">cd</span> node_modules\raylib
+<span class="linenos">2</span><span class="k">rmdir</span> /S /Q bin build
+<span class="linenos">3</span>npx cmake-js compile -t ClangCL
+</pre></div>
+</div>
 </section>
 </section>
 <section id="running-benchmarks">
 <h2>Running benchmarks<a class="headerlink" href="#running-benchmarks" title="Permalink to this heading">#</a></h2>
 <p>Open a console, go to <code class="docutils literal notranslate"><span class="pre">koffi/benchmark</span></code> and run <code class="docutils literal notranslate"><span class="pre">../../cnoke/cnoke.js</span></code> (or <code class="docutils literal notranslate"><span class="pre">node</span> <span class="pre">..\..\cnoke\cnoke.js</span></code> on Windows) before doing anything else.</p>
-<div class="highlight-sh notranslate"><div class="highlight"><pre><span></span><span class="nb">cd</span> koffi/benchmark
-node ../../cnoke/cnoke.js
+<p>Please note that all benchmark results are made with Clang-built binaries.</p>
+<div class="highlight-sh notranslate"><div class="highlight"><pre><span></span><span class="linenos">1</span><span class="nb">cd</span> koffi/benchmark
+<span class="linenos">2</span>node ../../cnoke/cnoke.js --prefer-clang
 </pre></div>
 </div>
-<p>Once this is done, you can execute each implementation, e.g. <code class="docutils literal notranslate"><span class="pre">build/raylib_cc</span></code> or <code class="docutils literal notranslate"><span class="pre">node</span> <span class="pre">./atoi_koffi.js</span></code>. You can optionally define a custom number of iterations, e.g. <code class="docutils literal notranslate"><span class="pre">node</span> <span class="pre">./atoi_koffi.js</span> <span class="pre">10000000</span></code>.</p>
-<div class="highlight-sh notranslate"><div class="highlight"><pre><span></span>node ./atoi_napi.js
-node ./atoi_koffi.js
+<p>Once everything is built and ready, run:</p>
+<div class="highlight-sh notranslate"><div class="highlight"><pre><span></span><span class="linenos">1</span>node benchmark.js
 </pre></div>
 </div>
 </section>
@@ -542,19 +533,17 @@ node ./atoi_koffi.js
           <div class="toc-tree">
             <ul>
 <li><a class="reference internal" href="#">Benchmarks</a><ul>
-<li><a class="reference internal" href="#rand-results">rand results</a><ul>
-<li><a class="reference internal" href="#linux-x86-64">Linux x86_64</a></li>
-<li><a class="reference internal" href="#windows-x86-64">Windows x86_64</a></li>
-</ul>
-</li>
-<li><a class="reference internal" href="#atoi-results">atoi results</a><ul>
-<li><a class="reference internal" href="#id1">Linux x86_64</a></li>
-<li><a class="reference internal" href="#id2">Windows x86_64</a></li>
+<li><a class="reference internal" href="#overview">Overview</a></li>
+<li><a class="reference internal" href="#linux-x86-64">Linux x86_64</a><ul>
+<li><a class="reference internal" href="#rand-results">rand results</a></li>
+<li><a class="reference internal" href="#atoi-results">atoi results</a></li>
+<li><a class="reference internal" href="#raylib-results">Raylib results</a></li>
 </ul>
 </li>
-<li><a class="reference internal" href="#raylib-results">Raylib results</a><ul>
-<li><a class="reference internal" href="#id3">Linux x86_64</a></li>
-<li><a class="reference internal" href="#id4">Windows x86_64</a></li>
+<li><a class="reference internal" href="#windows-x86-64">Windows x86_64</a><ul>
+<li><a class="reference internal" href="#id1">rand results</a></li>
+<li><a class="reference internal" href="#id2">atoi results</a></li>
+<li><a class="reference internal" href="#id3">Raylib results</a></li>
 </ul>
 </li>
 <li><a class="reference internal" href="#running-benchmarks">Running benchmarks</a></li>

package/doc/dist/html/changes.html

@@ -5,7 +5,7 @@
     <meta name="color-scheme" content="light dark"><meta name="generator" content="Docutils 0.18.1: http://docutils.sourceforge.net/" />
 <link rel="index" title="Index" href="genindex" /><link rel="search" title="Search" href="search" /><link rel="prev" title="Contributing" href="contribute" />
 
-    <meta name="generator" content="sphinx-5.0.1, furo 2022.06.04.1"/>
+    <meta name="generator" content="sphinx-5.0.2, furo 2022.06.21"/>
         <title>Changelog - Koffi</title>
       <link rel="stylesheet" type="text/css" href="_static/pygments.css" />
     <link rel="stylesheet" type="text/css" href="_static/styles/furo.css?digest=40978830699223671f4072448e654b5958f38b89" />
@@ -162,7 +162,7 @@
 </form>
 <div id="searchbox"></div><div class="sidebar-scroll"><div class="sidebar-tree">
   <ul class="current">
-<li class="toctree-l1"><a class="reference internal" href="platforms">Supported platforms</a></li>
+<li class="toctree-l1"><a class="reference internal" href="platforms">Requirements</a></li>
 <li class="toctree-l1"><a class="reference internal" href="start">Quick start</a></li>
 <li class="toctree-l1"><a class="reference internal" href="types">Data types</a></li>
 <li class="toctree-l1"><a class="reference internal" href="functions">Function calls</a></li>
@@ -205,6 +205,45 @@
         <article role="main">
           <section id="changelog">
 <h1>Changelog<a class="headerlink" href="#changelog" title="Permalink to this heading">#</a></h1>
+<section id="koffi-1-3-8">
+<h2>Koffi 1.3.8<a class="headerlink" href="#koffi-1-3-8" title="Permalink to this heading">#</a></h2>
+<p><strong>Main changes:</strong></p>
+<ul class="simple">
+<li><p>Prevent callback reuse beyond FFI call</p></li>
+<li><p>Add BTI support for AAarch64 platforms (except Windows)</p></li>
+</ul>
+<p><strong>Other changes:</strong></p>
+<ul class="simple">
+<li><p>Fix and harmonize a few error messages</p></li>
+</ul>
+</section>
+<section id="koffi-1-3-7">
+<h2>Koffi 1.3.7<a class="headerlink" href="#koffi-1-3-7" title="Permalink to this heading">#</a></h2>
+<p><strong>Main fixes:</strong></p>
+<ul class="simple">
+<li><p>Fix crash when using callbacks inside structs</p></li>
+<li><p>Support for null strings in record members</p></li>
+</ul>
+<p><strong>Other changes:</strong></p>
+<ul class="simple">
+<li><p>Add intptr_t and uintptr_t primitive types</p></li>
+<li><p>Add str/str16 type aliases for string/string16</p></li>
+<li><p>Various documentation fixes and improvements</p></li>
+</ul>
+</section>
+<section id="koffi-1-3-6">
+<h2>Koffi 1.3.6<a class="headerlink" href="#koffi-1-3-6" title="Permalink to this heading">#</a></h2>
+<p><strong>Main fixes:</strong></p>
+<ul class="simple">
+<li><p>Fix install error with Node &lt; 15 on Windows (build system bug)</p></li>
+</ul>
+<p><strong>Other changes:</strong></p>
+<ul class="simple">
+<li><p>Detect incompatible Node.js versions when installing Koffi</p></li>
+<li><p>Prebuild with Clang for Windows x64 and Linux x64 binaries</p></li>
+<li><p>Various documentation improvements</p></li>
+</ul>
+</section>
 <section id="koffi-1-3-5">
 <h2>Koffi 1.3.5<a class="headerlink" href="#koffi-1-3-5" title="Permalink to this heading">#</a></h2>
 <p><strong>Main changes:</strong></p>
@@ -361,6 +400,9 @@
           <div class="toc-tree">
             <ul>
 <li><a class="reference internal" href="#">Changelog</a><ul>
+<li><a class="reference internal" href="#koffi-1-3-8">Koffi 1.3.8</a></li>
+<li><a class="reference internal" href="#koffi-1-3-7">Koffi 1.3.7</a></li>
+<li><a class="reference internal" href="#koffi-1-3-6">Koffi 1.3.6</a></li>
 <li><a class="reference internal" href="#koffi-1-3-5">Koffi 1.3.5</a></li>
 <li><a class="reference internal" href="#koffi-1-3-4">Koffi 1.3.4</a></li>
 <li><a class="reference internal" href="#koffi-1-3-3">Koffi 1.3.3</a></li>