Sindbad~EG File Manager
<?xml version="1.0" encoding="UTF-8" standalone="no"?>
<!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">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8" />
<title>Dbstl persistence</title>
<link rel="stylesheet" href="gettingStarted.css" type="text/css" />
<meta name="generator" content="DocBook XSL Stylesheets V1.73.2" />
<link rel="start" href="index.html" title="Berkeley DB Programmer's Reference Guide" />
<link rel="up" href="stl.html" title="Chapter 7. Standard Template Library API" />
<link rel="prev" href="stl_complex_rw.html" title="Store and Retrieve data or objects of complex types" />
<link rel="next" href="stl_container_specific.html" title="Dbstl container specific notes" />
</head>
<body>
<div xmlns="" class="navheader">
<div class="libver">
<p>Library Version 18.1.40</p>
</div>
<table width="100%" summary="Navigation header">
<tr>
<th colspan="3" align="center">Dbstl persistence</th>
</tr>
<tr>
<td width="20%" align="left"><a accesskey="p" href="stl_complex_rw.html">Prev</a> </td>
<th width="60%" align="center">Chapter 7. Standard Template Library API</th>
<td width="20%" align="right"> <a accesskey="n" href="stl_container_specific.html">Next</a></td>
</tr>
</table>
<hr />
</div>
<div class="sect1" lang="en" xml:lang="en">
<div class="titlepage">
<div>
<div>
<h2 class="title" style="clear: both"><a id="stl_persistence"></a>Dbstl persistence</h2>
</div>
</div>
</div>
<div class="toc">
<dl>
<dt>
<span class="sect2">
<a href="stl_persistence.html#directdbget">Direct database get</a>
</span>
</dt>
<dt>
<span class="sect2">
<a href="stl_persistence.html#chg_persistence">Change persistence</a>
</span>
</dt>
<dt>
<span class="sect2">
<a href="stl_persistence.html#obj_life_persistence">Object life time and persistence </a>
</span>
</dt>
</dl>
</div>
<p>
The following sections provide information on how to
achieve persistence using dbstl.
</p>
<div class="sect2" lang="en" xml:lang="en">
<div class="titlepage">
<div>
<div>
<h3 class="title"><a id="directdbget"></a>Direct database get</h3>
</div>
</div>
</div>
<p>
Each container has a <span class="bold"><strong>begin()</strong></span> method
which produces an iterator.
These <span class="bold"><strong>begin</strong></span> methods take
a boolean parameter, <span class="bold"><strong>directdb_get</strong></span>, which
controls the caching behavior of the iterator. The default value of this
parameter is <code class="literal">true</code>.
</p>
<p>
If <span class="bold"><strong>directdb_get</strong></span> is
<code class="literal">true</code>, then the persistent object is
fetched anew from the database each time the iterator is
dereferenced as a pointer by use of the star-operator
(<span class="bold"><strong>*iterator</strong></span>) or by use
of the arrow-operator (<span class="bold"><strong>iterator->member</strong></span>).
If <span class="bold"><strong>directdb_get</strong></span> is
<code class="literal">false</code>, then the first dereferencing
of the iterator fetches the object from the database, but
later dereferences can return cached data.
</p>
<p>
With <span class="bold"><strong>directdb_get</strong></span> set
to <code class="literal">true</code>, if you call:
</p>
<pre class="programlisting">(*iterator).datamember1=new-value1;
(*iterator).datamember2=new-value2; </pre>
<p>
then the assignment to <code class="literal">datamember1</code>
will be lost, because the second dereferencing of the
iterator would cause the cached copy of the object to be
overwritten by the object's persistent data from the
database.
</p>
<p>
You also can use the arrow operator like this:
</p>
<pre class="programlisting">iterator->datamember1=new-value1;
iterator->datamember2=new-value2; </pre>
<p>
This works exactly the same way as <span class="bold"><strong>iterator::operator*</strong></span>. For this
reason, the same caching rules apply to arrow operators as
they do for star operators.
</p>
<p>
One way to avoid this problem is to create a reference
to the object, and use it to access the object:
</p>
<pre class="programlisting">container::value_type &ref = *iterator;
ref.datamember1=new-value1;
ref.datamember2=new-value2;
...// more member function calls and datamember assignments
ref._DB_STL_StoreElement(); </pre>
<p>
The above code will not lose the newly assigned value
of <code class="literal">ref.datamember1</code> in the way that the
previous example did.
</p>
<p>
In order to avoid these complications, you can assign
to the object referenced by an iterator with another
object of the same type like this:
</p>
<pre class="programlisting">container::value_type obj2;
obj2.datamember1 = new-value1;
obj2.datamember2 = new-value2;
*itr = obj2; </pre>
<p>
This code snippet causes the new values in
<code class="literal">obj2</code> to be stored into the
underlying database.
</p>
<p>
If you have two iterators going through the same
container like this:
</p>
<pre class="programlisting">for (iterator1 = v.begin(), iterator2 = v.begin();
iterator1 != v.end();
++iterator1, ++iterator2) {
*iterator1 = new_value;
print(*iterator2);
} </pre>
<p>
then the printed value will depend on the value of
<span class="bold"><strong>directdb_get</strong></span> with
which the iterator had been created. If <span class="bold"><strong>directdb_get</strong></span> is
<code class="literal">false</code>, then the original,
persistent value is printed; otherwise the newly assigned
value is returned from the cache when
<code class="literal">iterator2</code> is dereferenced. This
happens because each iterator has its own cached copy of
the persistent object, and the dereferencing of
<code class="literal">iterator2</code> refreshes
<code class="literal">iterator2</code>'s copy from the database,
retrieving the value stored by the assignment to
<code class="literal">*iterator1</code>.
</p>
<p>
Alternatively, you can set <span class="bold"><strong>directdb_get </strong></span> to
<code class="literal">false</code> and call <code class="methodname">iterator2->refresh()</code>
immediately before the dereferencing of
<code class="literal">iterator2</code>, so that
<code class="literal">iterator2</code>'s cached value is
refreshed.
</p>
<p>
If <span class="bold"><strong>directdb_get</strong></span> is
<code class="literal">false</code>, a few of the tests in
dbstl's test kit will fail. This is because the above
contrived case appears in several of C++ STL tests.
Consequently, the default value of the <span class="bold"><strong>directdb_get</strong></span> parameter in the
<code class="methodname">container::begin()</code> methods is
<code class="literal">true</code>. If your use cases avoid such
bizarre usage of iterators, you can set it to
<code class="literal">false</code>, which makes the iterator
read operation faster.
</p>
</div>
<div class="sect2" lang="en" xml:lang="en">
<div class="titlepage">
<div>
<div>
<h3 class="title"><a id="chg_persistence"></a>Change persistence</h3>
</div>
</div>
</div>
<p>
If you modify the object to which an iterator refers by
using one of the following:
</p>
<pre class="programlisting">(*iterator).member_function_call()</pre>
<p>
or
</p>
<pre class="programlisting">(*iterator).data_member = new_value</pre>
<p>
then you should call
<code class="methodname">iterator->_DB_STL_StoreElement()</code>
to store the change. Otherwise the change is lost after
the iterator moves on to other elements.
</p>
<p>
If you are storing a sequence, and you modified some
part of it, you should also call
<code class="methodname">iterator->_DB_STL_StoreElement()</code>
before moving the iterator.
</p>
<p>
And in both cases, if <span class="bold"><strong>directdb_get</strong></span> is
<code class="literal">true</code>
(this is the default value), you should call
<code class="methodname">_DB_STL_StoreElement()</code> after
the change and before the next iterator movement OR the
next dereferencing of the iterator by the star or arrow
operators (<code class="literal">iterator::operator*</code> or
<code class="literal">iterator::operator-></code>).
Otherwise, you will lose the change.
</p>
<p>
If you update the element by assigning to a
dereferenced iterator like this:
</p>
<pre class="programlisting">*iterator = new_element;</pre>
<p>
then you never have to call
<code class="methodname">_DB_STL_StoreElement()</code>
because the change is stored in the database
automatically.
</p>
</div>
<div class="sect2" lang="en" xml:lang="en">
<div class="titlepage">
<div>
<div>
<h3 class="title"><a id="obj_life_persistence"></a>Object life time and persistence </h3>
</div>
</div>
</div>
<p>
Dbstl is an interface to Berkeley DB, so it is used to
store data persistently. This is really a different
purpose from that of regular C++ STL. This difference in
their goals has implications on expected object lifetime:
In standard STL, when you store an object A of type ID
into C++ stl vector V using V.push_back(A), if a proper
copy constructor is provided in A's class type, then the
copy of A (call it B) and everything in B, such as another
object C pointed to by B's data member B.c_ptr, will be
stored in V and will live as long as B is still in V and V
is alive. B will be destroyed when V is destroyed or B is
erased from V.
</p>
<p>
This is not true for dbstl, which will copy A's data
and store it in the underlying database. The copy is by
default a shallow copy, but users can register their
object marshalling and unmarshalling functions using the
<code class="classname">DbstlElemTraits</code> class template.
So if A is passed to a <code class="classname">db_vector</code>
container, <code class="literal">dv</code>, by using
<code class="literal">dv.push_back(A)</code>, then dbstl copies
A's data using the registered functions, and stores data
into the underlying database. Consequently, A will be
valid, even if the container is destroyed, because it is
stored into the database.
</p>
<p>
If the copy is simply a shallow copy, and A is later
destroyed, then the pointer stored in the database will
become invalid. The next time we use the retrieved object,
we will be using an invalid pointer, which probably will
result in errors. To avoid this, store the referred object
C rather than the pointer member A.c_ptr itself, by
registering the right marshalling/unmarshalling function
with <code class="classname">DbstlElemTraits</code>.
</p>
<p>
For example, consider the following example class
declaration:
</p>
<pre class="programlisting">class ID
{
public:
string Name;
int Score;
}; </pre>
<p>
Here, the class ID has a data member <span class="bold"><strong>Name</strong></span>, which refers to a memory
address of the actual characters in the string. If we
simply shallow copy an object, <code class="literal">id</code>, of
class ID to store it, then the stored data,
<code class="literal">idd</code>, is invalid when
<code class="literal">id</code> is destroyed. This is because
<code class="literal">idd</code> and <code class="literal">id</code> refer
to a common memory address which is the base address of
the memory space storing all characters in the string, and
this memory space is released when <code class="literal">id</code>
is destroyed. So <code class="literal">idd</code> will be referring
to an invalid address. The next time we retrieve
<code class="literal">idd</code> and use it, there will probably
be memory corruption.
</p>
<p>
The way to store <code class="literal">id</code> is to write a
marshal/unmarshal function pair like this:
</p>
<pre class="programlisting">void copy_id(void *dest, const ID&elem)
{
memcpy(dest, &elem.Score, sizeof(elem.Score));
char *p = ((char *)dest) + sizeof(elem.Score);
strcpy(p, elem.Name.c_str());
}
void restore_id(ID& dest, const void *srcdata)
{
memcpy(&dest.Score, srcdata, sizeof(dest.Score));
const char *p = ((char *)srcdata) + sizeof(dest.Score);
dest.Name = p;
}
size_t size_id(const ID& elem)
{
return sizeof(elem.Score) + elem.Name.size() +
1;// store the '\0' char.
} </pre>
<p>
Then register the above functions before storing any
instance of <code class="classname">ID</code>:
</p>
<pre class="programlisting">DbstlElemTraits<ID>::instance()->set_copy_function(copy_id);
DbstlElemTraits<ID>::instance()->set_size_function(size_id);
DbstlElemTraits<ID>::instance()->set_restore_function(restore_id); </pre>
<p>
This way, the actual data of instances of ID are
stored, and so the data will persist even if the container
itself is destroyed.
</p>
</div>
</div>
<div class="navfooter">
<hr />
<table width="100%" summary="Navigation footer">
<tr>
<td width="40%" align="left"><a accesskey="p" href="stl_complex_rw.html">Prev</a> </td>
<td width="20%" align="center">
<a accesskey="u" href="stl.html">Up</a>
</td>
<td width="40%" align="right"> <a accesskey="n" href="stl_container_specific.html">Next</a></td>
</tr>
<tr>
<td width="40%" align="left" valign="top">Store and Retrieve data or
objects of complex types </td>
<td width="20%" align="center">
<a accesskey="h" href="index.html">Home</a>
</td>
<td width="40%" align="right" valign="top"> Dbstl container specific
notes</td>
</tr>
</table>
</div>
</body>
</html>
Sindbad File Manager Version 1.0, Coded By Sindbad EG ~ The Terrorists