boost_1_45_0/libs/python/doc/internals.html - nest-learning-thermostat/5.0/boost - Git at Google

 <?xml version="1.0" encoding="utf-8" ?>
 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
 <html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en" lang="en">
 <head>
 <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
 <meta name="generator" content="Docutils 0.3.0: http://docutils.sourceforge.net/" />
 <title>Boost.Python Internals Boost</title>
 <link rel="stylesheet" href="../../../rst.css" type="text/css" />
 </head>
 <body>
 <div class="document" id="boost-python-internals-logo">
 <h1 class="title"><a class="reference" href="index.html">Boost.Python</a> Internals <a class="reference" href="../../../index.htm"><img alt="Boost" src="../../../boost.png" /></a></h1>
 <div class="section" id="a-conversation-between-brett-calcott-and-david-abrahams">
 <h1><a name="a-conversation-between-brett-calcott-and-david-abrahams">A conversation between Brett Calcott and David Abrahams</a></h1>
 <table class="field-list" frame="void" rules="none">
 <col class="field-name" />
 <col class="field-body" />
 <tbody valign="top">
 <tr class="field"><th class="field-name">copyright:</th><td class="field-body">Copyright David Abrahams and Brett Calcott 2003. See
 accompanying <a class="reference" href="../../../LICENSE_1_0.txt">license</a> for terms of use.</td>
 </tr>
 </tbody>
 </table>
 <p>In both of these cases, I'm quite capable of reading code - but the
 thing I don't get from scanning the source is a sense of the
 architecture, both structurally, and temporally (er, I mean in what
 order things go on).</p>
 <ol class="arabic">
 <li><p class="first">What happens when you do the following:</p>
 <pre class="literal-block">
 struct boring {};
 ...etc...
 class_&lt;boring&gt;(&quot;boring&quot;)
     ;
 </pre>
 </li>
 </ol>
 <p>There seems to be a fair bit going on.</p>
 <blockquote>
 <ul class="simple">
 <li>Python needs a new ClassType to be registered.</li>
 <li>We need to construct a new type that can hold our boring struct.</li>
 <li>Inward and outward converters need to be registered for the type.</li>
 </ul>
 </blockquote>
 <p>Can you gesture in the general direction where these things are done?</p>
 <blockquote>
 <p>I only have time for a &quot;off-the-top-of-my-head&quot; answer at the moment;
 I suggest you step through the code with a debugger after reading this
 to see how it works, fill in details, and make sure I didn't forget
 anything.</p>
 <blockquote>
 <p>A new (Python) subclass of Boost.Python.Instance (see
 libs/python/src/object/class.cpp) is created by invoking
 Boost.Python.class, the metatype:</p>
 <pre class="literal-block">
 &gt;&gt;&gt; boring = Boost.Python.class(
 ...     'boring'
 ...   , bases_tuple       # in this case, just ()
 ...   , {
 ...         '__module__' : module_name
 ...       , '__doc__' : doc_string # optional
 ...     }
 ... )
 </pre>
 <p>A handle to this object is stuck in the m_class_object field
 of the registration associated with <tt class="literal"><span class="pre">typeid(boring)</span></tt>.  The
 registry will keep that object alive forever, even if you
 wipe out the 'boring' attribute of the extension module
 (probably not a good thing).</p>
 <p>Because you didn't specify <tt class="literal"><span class="pre">class&lt;boring,</span> <span class="pre">non_copyable,</span>
 <span class="pre">...&gt;</span></tt>, a to-python converter for boring is registered which
 copies its argument into a value_holder held by the the
 Python boring object.</p>
 <p>Because you didn't specify <tt class="literal"><span class="pre">class&lt;boring</span> <span class="pre">...&gt;(no_init)</span></tt>,
 an <tt class="literal"><span class="pre">__init__</span></tt> function object is added to the class
 dictionary which default-constructs a boring in a
 value_holder (because you didn't specify some smart pointer
 or derived wrapper class as a holder) held by the Python
 boring object.</p>
 <p><tt class="literal"><span class="pre">register_class_from_python</span></tt> is used to register a
 from-python converter for <tt class="literal"><span class="pre">shared_ptr&lt;boring&gt;</span></tt>.
 <tt class="literal"><span class="pre">boost::shared_ptr</span></tt>s are special among smart pointers
 because their Deleter argument can be made to manage the
 whole Python object, not just the C++ object it contains, no
 matter how the C++ object is held.</p>
 <p>If there were any <tt class="literal"><span class="pre">bases&lt;&gt;</span></tt>, we'd also be registering the
 relationship between these base classes and boring in the
 up/down cast graph (<tt class="literal"><span class="pre">inheritance.[hpp/cpp]</span></tt>).</p>
 <p>In earlier versions of the code, we'd be registering lvalue
 from-python converters for the class here, but now
 from-python conversion for wrapped classes is handled as a
 special case, before consulting the registry, if the source
 Python object's metaclass is the Boost.Python metaclass.</p>
 <p>Hmm, that from-python converter probably ought to be handled
 the way class converters are, with no explicit conversions
 registered.</p>
 </blockquote>
 </blockquote>
 <ol class="arabic" start="2">
 <li><p class="first">Can you give a brief overview of the data structures that are
 present in the registry</p>
 <blockquote>
 <p>The registry is simple: it's just a map from typeid -&gt;
 registration (see boost/python/converter/registrations.hpp).
 <tt class="literal"><span class="pre">lvalue_chain</span></tt> and <tt class="literal"><span class="pre">rvalue_chain</span></tt> are simple endogenous
 linked lists.</p>
 <p>If you want to know more, just ask.</p>
 <p>If you want to know about the cast graph, ask me something specific in
 a separate message.</p>
 </blockquote>
 <p>and an overview of the process that happens as a type makes its
 way from c++ to python and back again.</p>
 </li>
 </ol>
 <blockquote>
 <p>Big subject.  I suggest some background reading: look for relevant
 info in the LLNL progress reports and the messages they link to.
 Also,</p>
 <blockquote>
 <p><a class="reference" href="http://mail.python.org/pipermail/c++-sig/2002-May/001023.html">http://mail.python.org/pipermail/c++-sig/2002-May/001023.html</a></p>
 <p><a class="reference" href="http://mail.python.org/pipermail/c++-sig/2002-December/003115.html">http://mail.python.org/pipermail/c++-sig/2002-December/003115.html</a></p>
 <p><a class="reference" href="http://aspn.activestate.com/ASPN/Mail/Message/1280898">http://aspn.activestate.com/ASPN/Mail/Message/1280898</a></p>
 <p><a class="reference" href="http://mail.python.org/pipermail/c++-sig/2002-July/001755.html">http://mail.python.org/pipermail/c++-sig/2002-July/001755.html</a></p>
 </blockquote>
 <p>from c++ to python:</p>
 <blockquote>
 <p>It depends on the type and the call policies in use or, for
 <tt class="literal"><span class="pre">call&lt;&gt;(...)</span></tt>, <tt class="literal"><span class="pre">call_method&lt;&gt;(...)</span></tt>, or <tt class="literal"><span class="pre">object(...)</span></tt>, if
 <tt class="literal"><span class="pre">ref</span></tt> or <tt class="literal"><span class="pre">ptr</span></tt> is used.  There are also two basic
 categories to to-python conversion, &quot;return value&quot; conversion
 (for Python-&gt;C++ calls) and &quot;argument&quot; conversion (for
 C++-&gt;Python calls and explicit <tt class="literal"><span class="pre">object()</span></tt> conversions).  The
 behavior of these two categories differs subtly in various ways
 whose details I forget at the moment.  You can probably find
 the answers in the above references, and certainly in the code.</p>
 <p>The &quot;default&quot; case is by-value (copying) conversion, which uses
 to_python_value as a to-python converter.</p>
 <blockquote>
 <p>Since there can sensibly be only one way to convert any type
 to python (disregarding the idea of scoped registries for the
 moment), it makes sense that to-python conversions can be
 handled by specializing a template.  If the type is one of
 the types handled by a built-in conversion
 (builtin_converters.hpp), the corresponding template
 specialization of to_python_value gets used.</p>
 <p>Otherwise, to_python_value uses the <tt class="literal"><span class="pre">m_to_python</span></tt>
 function in the registration for the C++ type.</p>
 </blockquote>
 <p>Other conversions, like by-reference conversions, are only
 available for wrapped classes, and are requested explicitly by
 using <tt class="literal"><span class="pre">ref(...)</span></tt>, <tt class="literal"><span class="pre">ptr(...)</span></tt>, or by specifying different
 CallPolicies for a call, which can cause a different to-python
 converter to be used.  These conversions are never registered
 anywhere, though they do need to use the registration to find
 the Python class corresponding to the C++ type being referred
 to.  They just build a new Python instance and stick the
 appropriate Holder instance in it.</p>
 </blockquote>
 <p>from python to C++:</p>
 <blockquote>
 <p>Once again I think there is a distinction between &quot;return value&quot;
 and &quot;argument&quot; conversions, and I forget exactly what that is.</p>
 <p>What happens depends on whether an lvalue conversion is needed
 (see <a class="reference" href="http://mail.python.org/pipermail/c++-sig/2002-May/001023.html">http://mail.python.org/pipermail/c++-sig/2002-May/001023.html</a>)
 All lvalue conversions are also registered in a type's rvalue
 conversion chain, since when an rvalue will do, an lvalue is
 certainly good enough.</p>
 <p>An lvalue conversion can be done in one step (just get me the
 pointer to the object - it can be <tt class="literal"><span class="pre">NULL</span></tt> if no conversion is
 possible) while an rvalue conversion requires two steps to
 support wrapped function overloading and multiple converters for
 a given C++ target type: first tell me if a conversion is
 possible, then construct the converted object as a second step.</p>
 </blockquote>
 </blockquote>
 </div>
 </div>
 <hr class="footer"/>
 <div class="footer">
 <a class="reference" href="internals.rst">View document source</a>.
 Generated on: 2003-09-12 14:51 UTC.
 Generated by <a class="reference" href="http://docutils.sourceforge.net/">Docutils</a> from <a class="reference" href="http://docutils.sourceforge.net/rst.html">reStructuredText</a> source.
 </div>
 </body>
 </html>
	<?xml version="1.0" encoding="utf-8" ?>
	<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
	<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en" lang="en">
	<head>
	<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
	<meta name="generator" content="Docutils 0.3.0: http://docutils.sourceforge.net/" />
	<title>Boost.Python Internals Boost</title>
	<link rel="stylesheet" href="../../../rst.css" type="text/css" />
	</head>
	<body>
	<div class="document" id="boost-python-internals-logo">
	<h1 class="title"><a class="reference" href="index.html">Boost.Python</a> Internals <a class="reference" href="../../../index.htm"><img alt="Boost" src="../../../boost.png" /></a></h1>
	<div class="section" id="a-conversation-between-brett-calcott-and-david-abrahams">
	<h1><a name="a-conversation-between-brett-calcott-and-david-abrahams">A conversation between Brett Calcott and David Abrahams</a></h1>
	<table class="field-list" frame="void" rules="none">
	<col class="field-name" />
	<col class="field-body" />
	<tbody valign="top">
	<tr class="field"><th class="field-name">copyright:</th><td class="field-body">Copyright David Abrahams and Brett Calcott 2003. See
	accompanying <a class="reference" href="../../../LICENSE_1_0.txt">license</a> for terms of use.</td>
	</tr>
	</tbody>
	</table>
	<p>In both of these cases, I'm quite capable of reading code - but the
	thing I don't get from scanning the source is a sense of the
	architecture, both structurally, and temporally (er, I mean in what
	order things go on).</p>
	<ol class="arabic">
	<li><p class="first">What happens when you do the following:</p>
	<pre class="literal-block">
	struct boring {};
	...etc...
	class_<boring>("boring")
	;
	</pre>
	</li>
	</ol>
	<p>There seems to be a fair bit going on.</p>
	<blockquote>
	<ul class="simple">
	<li>Python needs a new ClassType to be registered.</li>
	<li>We need to construct a new type that can hold our boring struct.</li>
	<li>Inward and outward converters need to be registered for the type.</li>
	</ul>
	</blockquote>
	<p>Can you gesture in the general direction where these things are done?</p>
	<blockquote>
	<p>I only have time for a "off-the-top-of-my-head" answer at the moment;
	I suggest you step through the code with a debugger after reading this
	to see how it works, fill in details, and make sure I didn't forget
	anything.</p>
	<blockquote>
	<p>A new (Python) subclass of Boost.Python.Instance (see
	libs/python/src/object/class.cpp) is created by invoking
	Boost.Python.class, the metatype:</p>
	<pre class="literal-block">
	>>> boring = Boost.Python.class(
	... 'boring'
	... , bases_tuple # in this case, just ()
	... , {
	... '__module__' : module_name
	... , '__doc__' : doc_string # optional
	... }
	... )
	</pre>
	<p>A handle to this object is stuck in the m_class_object field
	of the registration associated with <tt class="literal"><span class="pre">typeid(boring)</span></tt>. The
	registry will keep that object alive forever, even if you
	wipe out the 'boring' attribute of the extension module
	(probably not a good thing).</p>
	<p>Because you didn't specify <tt class="literal"><span class="pre">class<boring,</span> <span class="pre">non_copyable,</span>
	<span class="pre">...></span></tt>, a to-python converter for boring is registered which
	copies its argument into a value_holder held by the the
	Python boring object.</p>
	<p>Because you didn't specify <tt class="literal"><span class="pre">class<boring</span> <span class="pre">...>(no_init)</span></tt>,
	an <tt class="literal"><span class="pre">__init__</span></tt> function object is added to the class
	dictionary which default-constructs a boring in a
	value_holder (because you didn't specify some smart pointer
	or derived wrapper class as a holder) held by the Python
	boring object.</p>
	<p><tt class="literal"><span class="pre">register_class_from_python</span></tt> is used to register a
	from-python converter for <tt class="literal"><span class="pre">shared_ptr<boring></span></tt>.
	<tt class="literal"><span class="pre">boost::shared_ptr</span></tt>s are special among smart pointers
	because their Deleter argument can be made to manage the
	whole Python object, not just the C++ object it contains, no
	matter how the C++ object is held.</p>
	<p>If there were any <tt class="literal"><span class="pre">bases<></span></tt>, we'd also be registering the
	relationship between these base classes and boring in the
	up/down cast graph (<tt class="literal"><span class="pre">inheritance.[hpp/cpp]</span></tt>).</p>
	<p>In earlier versions of the code, we'd be registering lvalue
	from-python converters for the class here, but now
	from-python conversion for wrapped classes is handled as a
	special case, before consulting the registry, if the source
	Python object's metaclass is the Boost.Python metaclass.</p>
	<p>Hmm, that from-python converter probably ought to be handled
	the way class converters are, with no explicit conversions
	registered.</p>
	</blockquote>
	</blockquote>
	<ol class="arabic" start="2">
	<li><p class="first">Can you give a brief overview of the data structures that are
	present in the registry</p>
	<blockquote>
	<p>The registry is simple: it's just a map from typeid ->
	registration (see boost/python/converter/registrations.hpp).
	<tt class="literal"><span class="pre">lvalue_chain</span></tt> and <tt class="literal"><span class="pre">rvalue_chain</span></tt> are simple endogenous
	linked lists.</p>
	<p>If you want to know more, just ask.</p>
	<p>If you want to know about the cast graph, ask me something specific in
	a separate message.</p>
	</blockquote>
	<p>and an overview of the process that happens as a type makes its
	way from c++ to python and back again.</p>
	</li>
	</ol>
	<blockquote>
	<p>Big subject. I suggest some background reading: look for relevant
	info in the LLNL progress reports and the messages they link to.
	Also,</p>
	<blockquote>
	<p><a class="reference" href="http://mail.python.org/pipermail/c++-sig/2002-May/001023.html">http://mail.python.org/pipermail/c++-sig/2002-May/001023.html</a></p>
	<p><a class="reference" href="http://mail.python.org/pipermail/c++-sig/2002-December/003115.html">http://mail.python.org/pipermail/c++-sig/2002-December/003115.html</a></p>
	<p><a class="reference" href="http://aspn.activestate.com/ASPN/Mail/Message/1280898">http://aspn.activestate.com/ASPN/Mail/Message/1280898</a></p>
	<p><a class="reference" href="http://mail.python.org/pipermail/c++-sig/2002-July/001755.html">http://mail.python.org/pipermail/c++-sig/2002-July/001755.html</a></p>
	</blockquote>
	<p>from c++ to python:</p>
	<blockquote>
	<p>It depends on the type and the call policies in use or, for
	<tt class="literal"><span class="pre">call<>(...)</span></tt>, <tt class="literal"><span class="pre">call_method<>(...)</span></tt>, or <tt class="literal"><span class="pre">object(...)</span></tt>, if
	<tt class="literal"><span class="pre">ref</span></tt> or <tt class="literal"><span class="pre">ptr</span></tt> is used. There are also two basic
	categories to to-python conversion, "return value" conversion
	(for Python->C++ calls) and "argument" conversion (for
	C++->Python calls and explicit <tt class="literal"><span class="pre">object()</span></tt> conversions). The
	behavior of these two categories differs subtly in various ways
	whose details I forget at the moment. You can probably find
	the answers in the above references, and certainly in the code.</p>
	<p>The "default" case is by-value (copying) conversion, which uses
	to_python_value as a to-python converter.</p>
	<blockquote>
	<p>Since there can sensibly be only one way to convert any type
	to python (disregarding the idea of scoped registries for the
	moment), it makes sense that to-python conversions can be
	handled by specializing a template. If the type is one of
	the types handled by a built-in conversion
	(builtin_converters.hpp), the corresponding template
	specialization of to_python_value gets used.</p>
	<p>Otherwise, to_python_value uses the <tt class="literal"><span class="pre">m_to_python</span></tt>
	function in the registration for the C++ type.</p>
	</blockquote>
	<p>Other conversions, like by-reference conversions, are only
	available for wrapped classes, and are requested explicitly by
	using <tt class="literal"><span class="pre">ref(...)</span></tt>, <tt class="literal"><span class="pre">ptr(...)</span></tt>, or by specifying different
	CallPolicies for a call, which can cause a different to-python
	converter to be used. These conversions are never registered
	anywhere, though they do need to use the registration to find
	the Python class corresponding to the C++ type being referred
	to. They just build a new Python instance and stick the
	appropriate Holder instance in it.</p>
	</blockquote>
	<p>from python to C++:</p>
	<blockquote>
	<p>Once again I think there is a distinction between "return value"
	and "argument" conversions, and I forget exactly what that is.</p>
	<p>What happens depends on whether an lvalue conversion is needed
	(see <a class="reference" href="http://mail.python.org/pipermail/c++-sig/2002-May/001023.html">http://mail.python.org/pipermail/c++-sig/2002-May/001023.html</a>)
	All lvalue conversions are also registered in a type's rvalue
	conversion chain, since when an rvalue will do, an lvalue is
	certainly good enough.</p>
	<p>An lvalue conversion can be done in one step (just get me the
	pointer to the object - it can be <tt class="literal"><span class="pre">NULL</span></tt> if no conversion is
	possible) while an rvalue conversion requires two steps to
	support wrapped function overloading and multiple converters for
	a given C++ target type: first tell me if a conversion is
	possible, then construct the converted object as a second step.</p>
	</blockquote>
	</blockquote>
	</div>
	</div>
	<hr class="footer"/>
	<div class="footer">
	<a class="reference" href="internals.rst">View document source</a>.
	Generated on: 2003-09-12 14:51 UTC.
	Generated by <a class="reference" href="http://docutils.sourceforge.net/">Docutils</a> from <a class="reference" href="http://docutils.sourceforge.net/rst.html">reStructuredText</a> source.
	</div>
	</body>
	</html>