svn: /phpdoc/en/trunk/reference/misc/functions/ pack.xml

From: Date: Sun, 05 Sep 2010 09:59:08 +0000
Subject: svn: /phpdoc/en/trunk/reference/misc/functions/ pack.xml
Groups: php.doc.cvs 
Request: Send a blank email to doc-cvs+get-7060@lists.php.net to get a copy of this message
cataphract Sun, 05 Sep 2010 09:59:08 +0000 Revision: http://svn.php.net/viewvc?view=revision&revision=303052 Log: - Expanding on the large integers values and unsigned types. Tangentially related to bug #52729. Bug: http://bugs.php.net/52729 (Open) unpack() format I, L, N and V returns negative value Changed paths: U phpdoc/en/trunk/reference/misc/functions/pack.xml Modified: phpdoc/en/trunk/reference/misc/functions/pack.xml =================================================================== --- phpdoc/en/trunk/reference/misc/functions/pack.xml 2010-09-05 09:29:37 UTC (rev 303051) +++ phpdoc/en/trunk/reference/misc/functions/pack.xml 2010-09-05 09:59:08 UTC (rev 303052) @@ -5,7 +5,7 @@ <refname>pack</refname> <refpurpose>Pack data into binary string</refpurpose> </refnamediv> - + <refsect1 role="description"> &reftitle.description; <methodsynopsis> @@ -29,14 +29,8 @@ function <function>pack</function> gives the same result for signed and unsigned format codes. </para> - <para> - Also note that PHP internally stores <type>integer</type> values as - signed values of a machine-dependent size. If you give it an unsigned - integer value too large to be stored that way it is converted to a - <type>float</type> which often yields an undesired result. - </para> </refsect1> - + <refsect1 role="parameters"> &reftitle.parameters; <para> @@ -159,14 +153,14 @@ </variablelist> </para> </refsect1> - + <refsect1 role="returnvalues"> &reftitle.returnvalues; <para> Returns a binary string containing data. </para> </refsect1> - + <refsect1 role="examples"> &reftitle.examples; <para> @@ -186,7 +180,34 @@ </example> </para> </refsect1> - + + <refsect1 role="notes"> + &reftitle.notes; + <caution> + <para>Note that PHP internally stores <type>integer</type> values as + signed values of a machine-dependent size (C type <literal>long</literal>). + Integer literals and operations that yield numbers outside the bounds of the + <type>integer</type> type will be stored as <type>float</type>. + When packing these floats as integers, they are first cast into the integer + type. This may or may not result in the desired byte pattern. + </para> + <para>The most relevant case is when packing unsigned numbers that would + be representable with the <type>integer</type> type if it were unsigned. + In systems where the <type>integer</type> type has a 32-bit size, the cast + usually results in the same byte pattern as if the <type>integer</type> were + unsigned (although this relies on implementation-defined unsigned to signed + conversions, as per the C standard). In systems where the + <type>integer</type> type has 64-bit size, the <type>float</type> most + likely does not have a mantissa large enough to hold the value without + loss of precision. If those systems also have a native 64-bit C + <literal>int</literal> type (most UNIX-like systems don't), the only way to + use the <literal>I</literal> pack format in the upper range is to create + <type>integer</type> negative values with the same byte representation + as the desired unsigned value. + </para> + </caution> + </refsect1> + <refsect1 role="seealso"> &reftitle.seealso; <para> @@ -195,7 +216,7 @@ </simplelist> </para> </refsect1> - + </refentry> <!-- Keep this comment at the end of the file

« previous php.doc.cvs (#7060) next »