blob: 789ebc16356f2c34b527adc9493f7dfbb71a5c09 [file] [log] [blame]
<!doctype HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<html>
<!--
(C) Copyright 2002-4 Robert Ramey - http://www.rrsd.com .
Use, modification and distribution is subject to the Boost Software
License, Version 1.0. (See accompanying file LICENSE_1_0.txt or copy at
http://www.boost.org/LICENSE_1_0.txt)
-->
<head>
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
<link rel="stylesheet" type="text/css" href="../../../boost.css">
<link rel="stylesheet" type="text/css" href="style.css">
<title>Serialization - Archive Exceptions</title>
</head>
<body link="#0000ff" vlink="#800080">
<table border="0" cellpadding="7" cellspacing="0" width="100%" summary="header">
<tr>
<td valign="top" width="300">
<h3><a href="../../../index.htm"><img height="86" width="277" alt="C++ Boost" src="../../../boost.png" border="0"></a></h3>
</td>
<td valign="top">
<h1 align="center">Serialization</h1>
<h2 align="center">Archive Exceptions</h2>
</td>
</tr>
</table>
<hr>
<dl class="page-index">
<dt><a href="#unregistered_class"><code style="white-space: normal">unregistered_class</code></a>
<dt><a href="#invalid_signature"><code style="white-space: normal">invalid_signature</code></a>
<dt><a href="#unsupported_version"><code style="white-space: normal">unsupported_version</code></a>
<dt><a href="#unsupported_class_version"><code style="white-space: normal">unsupported_class_version</code></a>
<dt><a href="#pointer_conflict"><code style="white-space: normal">pointer_conflict</code></a>
<dt><a href="#incompatible_native_format"><code style="white-space: normal">incompatible_native_format</code></a>
<dt><a href="#array_size_too_short"><code style="white-space: normal">array_size_too_short</code></a>
<dt><a href="#input_stream_error"><code style="white-space: normal">input_stream_error</code></a>
<dt><a href="#output_stream_error"><code style="white-space: normal">output_stream_error</code></a>
<dt><a href="#invalid_class_name"><code style="white-space: normal">invalid_class_name</code></a>
<dt><a href="#unregistered_class"><code style="white-space: normal">unregistered_class</code></a>
<dt><a href="#multiple_code_instantiation"><code style="white-space: normal">multiple_code_instantiation</code></a>
<dt><a href="#xml_archive_parsing_error"><code style="white-space: normal">xml_archive_parsing_error</code></a>
<dt><a href="#xml_archive_tag_mismatch"><code style="white-space: normal">xml_archive_tag_mismatch</code></a>
<dt><a href="#xml_archive_tag_name_error"><code style="white-space: normal">xml_archive_tag_name_error</code></a>
</dl>
Archive operators can throw a <code style="white-space: normal">boost::archive_exception</code>
object which can be caught by an application program. These exceptions are defined
in the files <a target="archive_exception_hpp" href="../../../boost/archive/archive_exception.hpp">
archive_exception.hpp</a>
and <a target="basic_xml_archive_hpp" href="../../../boost/archive/basic_xml_archive.hpp">
basic_xml_archive.hpp</a>.
<pre><code>
namespace boost {
namespace archive {
class archive_exception : public std::exception
{
public:
typedef enum {
unregistered_class, // attempt to serialize a pointer of an
// an unregistered class
invalid_signature, // first line of archive does not contain
// expected string
unsupported_version, // archive created with library version subsequent
// to this one
pointer_conflict // an attempt has been made to directly serialize
// an object after having already serialized the same
// object through a pointer. Were this permitted,
// it the archive load would result in the creation
// of extraneous object.
incompatible_native_format, // attempt to read native binary format
// on incompatible platform
array_size_too_short, // array being loaded doesn't fit in array allocated
input_stream_error // error on stream input
invalid_class_name, // class name greater than the maximum permitted.
// most likely a corrupted archive or an attempt
// to insert virus via buffer overrun method.
unregistered_cast, // base - derived relationship not registered with
// void_cast_register
unsupported_class_version, // type saved with a version # greater than the
// one used by the program. This indicates that the proggram
// needs to be rebuilt.
multiple_code_instantiation, // code for implementing serialization for some
// type has been instantiated in more than one module.
output_stream_error // error on stream output
} exception_code;
exception_code code;
archive_exception(exception_code c) : code(c) {}
virtual const char *what( ) const throw();
};
class xml_archive_exception : public virtual archive_exception
{
public:
typedef enum {
xml_archive_parsing_error, // archive doesn't contain expected data
xml_archive_tag_mismatch, // start/end tag in archive doesn't match program
xml_archive_tag_name_error // tag name contains invalid characters
} exception_code;
xml_archive_exception(exception_code c){}
virtual const char *what( ) const throw();
};
} // archive
} // boost
</code></pre>
<p>
<h3><a name="unregistered_class"><code style="white-space: normal">unregistered_class</code></a></h3>
An attempt has been made to serialize a polymorphic class through a pointer
without either registering it or associating it with an export key. This can also occur
when using a new archive how class name has not been added to the system with the
<code style="white-space: normal">BOOST_ARCHIVE_CUSTOM_ARCHIVE_TYPES</code> macro.
<h3><a name="invalid_signature"><code style="white-space: normal">invalid_signature</code></a></h3>
Archives are initiated with a known string. If this string is not found when
the archive is opened, It is presumed that this file is not a valid archive and this
exception is thrown.
<h3><a name="unsupported_version"><code style="white-space: normal">unsupported_version</code></a></h3>
This system assigns a version number of 3 to all archives created. Note that this is in
no way related to version number of classes used by application programs. This refers
to the version of the serialization system used to create the archive. Future versions
of this serialization system will be able to identify archives created under previous
(i.e. this) system and alter the loading procedure accordingly. Hence, future enhancements
to this serialization system should not obsolete any existing archive files. It is only
necessary to increment this version number when the newer system creates archives
incompatible in format with the current one.
<p>Should it ever occur that an older program attempts to read newer archives whose
format has changed, this exception is thrown.
<h3><a name="unsupported_class_version"><code style="white-space: normal">unsupported_class_version</code></a></h3>
An attempt has been made to load a class whose version has been incremented since the
program was written. Suppose that a class has been assigned version number 3 and the program
has been built and sent to third parties. Now suppose that the definition of that class
has been altered, the version number has been incremented to 4 and new archives have been
built. If one attempts to load these new archives with the original program, this
exception will be thrown.
<h3><a name="pointer_conflict"><code style="white-space: normal">pointer_conflict</code></a></h3>
To understand what this exception means consider the following scenario
<pre><code>
template&lt;class Archive&gt;
void T::save(Archive &ar) const
{
const A * aptr = &a;
ar << aptr; // save an instance of object of class A through a pointer
...
ar << a; // save an instance of an object of class A
assert(aptr == &a); // this must be true
}
template&lt;class Archive&gt;
void T::load(Archive &ar)
{
A * aptr;
ar >> aptr; // create and initialize a new instance of class A
...
ar >> a; // restore state of on object of class A
assert(aptr == &a); // this won't be true
}
</pre></code>
An object is saved first through a pointer then directly. Upon loading back
in the same sequence, we first create an new object and load in its data. Then
we load the data into another existing object. Where we started with one
object during save, we have two objects after restore. In a more realistic
situation, it could be very difficult to find this error. Fortunately,
these situations can be detected when the archive is created. When
this occurs, this exception is thrown.
<h3><a name = "incompatible_native_format"><code style="white-space: normal">incompatible_native_format</code></a></h3>
The library currently supports char text, wide char text and native binary
archive files. At the beginning of every archive, a signature is written indicating
the type of archive. This exception is thrown when an attempt is made to read
an archive written in a different format.
<h3><a name="array_size_too_short"><code style="white-space: normal">array_size_too_short</code></a></h3>
An attempt has been made to read an array that is larger than the array size.
This should only occur when the size of an array in code is reduced after an
archive has already been created.
<h3>
<a name="input_stream_error"><code style="white-space: normal">input_stream_error</code></a>
<br>
<a name="output_stream_error"><code style="white-space: normal">output_stream_error</code></a>
</h3>
An error has occured during stream input or ouput. Aside from the common
situations such as a corrupted or truncated input file, there are
several less obvious ones that sometimes occur.
<p>
This includes
an attempt to read past the end of the file. Text files need a terminating
new line character at the end of the file which is appended when the
archive destructor is invoked. Be sure that an output archive on a stream
is destroyed before opening an input archive on that same stream. That is,
rather than using something like:
<pre><code>
std::stringstream ss;
std::vector&lt;V&gt; v;
boost::archive::text_oarchive oa(ss);
oa &lt;&lt; v;
boost::archive::text_iarchive ia(ss);
ia &gt;&gt; v;
</code></pre>
use
<pre><code>
std::stringstream ss;
std::vector&lt;V&gt; v;
{
boost::archive::text_oarchive oa(ss);
oa &lt;&lt; v;
}
{
boost::archive::text_iarchive ia(ss);
ia &gt;&gt; v;
}
</code></pre>
<p>
Another one is the passing of uninitialized data. In general, the behavior
of the serialization library when passed uninitialized data is undefined.
If it can be detected, it will invoke an assertion in debug builds.
Otherwise, depending on the type of archive, it may pass through without
incident or it may result in an archive with unexpected data in it.
This, in turn, can result in the throwing of this exception.
<h3><a name="invalid_class_name"><code style="white-space: normal">invalid_class_name</code></a></h3>
Class name length greater than the maximum permitted. Most likely cause is a corrupted
archive or an attempt to insert virus via buffer overrun method.
<h3><a name="unregistered_cast"><code style="white-space: normal">unregistered_cast</code></a></h3>
In order to support casting between pointers of base and derived classes
at runtime, a collection of legitimate conversions is maintained by the system.
Normally this collection is maintained without any explicit action
on the part of the user of the library. However, there are special cases
where this might have to be done explicitly and could be overlooked. This
is described in <a href="serialization.html#runtimecasting">Runtime Casting</a>.
This exception is thrown if an attempt is made to convert between two pointers
whose relationship has not been registered,
<h3><a name="multiple_code_instantiation"><code style="white-space: normal">multiple_code_instantiation</code></a></h3>
This exception is thrown when it is detected that the serialization of the same type
has been instantiated more that once. This might occur when
serialization code is instantiated in both the mainline and one or more DLLS.
<h3><a name="xml_archive_parsing_error"><code style="white-space: normal">xml_archive_parsing_error</code></a></h3>
The XML generated by the serialization process is intimately coupled to the
C++ class structure, relationships between objects and the serialization
specifications. If these become out of sync in any way, the XML may not map
to the loading serialization and this exception might be thrown. This might
occur for one of the following reasons:
<ul>
<li>The archive has been edited outside the serialization system. This might
be possible if only the data is changed and not the XML attributes and nesting
structure is left unaltered. But any other editing is likely to render the
archive unreadable by the serialization library.
<li>The serialization has been altered and an archive generated by the old
code is being read. That is, versioning has not been properly employed to
properly deserialize previously created archives.
</ul>
<h3><a name="xml_archive_tag_mismatch"><code style="white-space: normal">xml_archive_tag_mismatch</code></a></h3>
This exception will be thrown if the start or end tag of and XML element doesn't match
the name specified for the object in the program.
<h3><a name="xml_archive_tag_name_error"><code style="white-space: normal">xml_archive_tag_name_error</code></a></h3>
This exception will be thrown if the tag name contains invalid characters. Valid characters
for an XML tag are: upper and lower case letters, digits, and the following punctuation: .(period),
_(underscore), :(colon), and -(hyphen).
<hr>
<p><i>&copy; Copyright <a href="http://www.rrsd.com">Robert Ramey</a> 2002-2004.
Distributed under the Boost Software License, Version 1.0. (See
accompanying file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
</i></p>
</body>
</html>