repo_zf1/documentation/manual/en/ref/migration-110.xml

<?xml version="1.0" encoding="UTF-8"?>
<!-- Reviewed: no -->
<sect1 id="migration.110">
    <title>Zend Framework 1.10</title>

    <para>
        When upgrading from a previous release to Zend Framework 1.10 or higher you
        should note the following migration notes.
    </para>

    <sect2 id="migration.110.zend.controller.front">
        <title>Zend_Controller_Front</title>

        <para>
            A wrong behaviour was fixed, when there was no module route and no route
            matched the given request. Previously, the router returned an unmodified
            request object, so the front controller just displayed the default controller
            and action. Since Zend Framework 1.10, the router will correctly as noted
            in the router interface, throw an exception if no route matches. The error
            plugin will then catch that exception and forward to the error controller.
            You can then test for that specific error with the constant
            <constant>Zend_Controller_Plugin_ErrorHandler::EXCEPTION_NO_ROUTE</constant>:
        </para>

        <programlisting language="php"><![CDATA[
/**
 * Before 1.10
 */
    public function errorAction()
    {
        $errors = $this->_getParam('error_handler');

        switch ($errors->type) {
            case Zend_Controller_Plugin_ErrorHandler::EXCEPTION_NO_CONTROLLER:
            case Zend_Controller_Plugin_ErrorHandler::EXCEPTION_NO_ACTION:
    // ...

/**
 * With 1.10
 */
    public function errorAction()
    {
        $errors = $this->_getParam('error_handler');

        switch ($errors->type) {
            case Zend_Controller_Plugin_ErrorHandler::EXCEPTION_NO_ROUTE:
            case Zend_Controller_Plugin_ErrorHandler::EXCEPTION_NO_CONTROLLER:
            case Zend_Controller_Plugin_ErrorHandler::EXCEPTION_NO_ACTION:
    // ...
]]></programlisting>
    </sect2>

    <sect2 id="migration.110.zend.feed.reader">
        <title>Zend_Feed_Reader</title>

        <para>
            With the introduction of Zend Framework 1.10, <classname>Zend_Feed_Reader</classname>'s
            handling of retrieving Authors and Contributors was changed, introducing
            a break in backwards compatibility. This change was an effort to harmonise
            the treatment of such data across the RSS and Atom classes of the component
            and enable the return of Author and Contributor data in more accessible,
            usable and detailed form. It also rectifies an error in that it was assumed
            any author element referred to a name. In RSS this is incorrect as an
            author element is actually only required to provide an email address.
            In addition, the original implementation applied its RSS limits to Atom
            feeds significantly reducing the usefulness of the parser with that format.
        </para>

        <para>
            The change means that methods like <methodname>getAuthors()</methodname>
            and <methodname>getContributors</methodname> no longer return a simple array
            of strings parsed from the relevant RSS and Atom elements. Instead, the return
            value is an <classname>ArrayObject</classname> subclass called
            <classname>Zend_Feed_Reader_Collection_Author</classname> which simulates
            an iterable multidimensional array of Authors. Each member of this object
            will be a simple array with three potential keys (as the source data permits).
            These include: name, email and uri.
        </para>

        <para>
            The original behaviour of such methods would have returned a simple
            array of strings, each string attempting to present a single name, but
            in reality this was unreliable since there is no rule governing the format
            of RSS Author strings.
        </para>

        <para>
            The simplest method of simulating the original behaviour of these
            methods is to use the <classname>Zend_Feed_Reader_Collection_Author</classname>'s
            <methodname>getValues()</methodname> which also returns a simple array of strings
            representing the "most relevant data", for authors presumed to be their name.
            Each value in the resulting array is derived from the "name" value
            attached to each Author (if present). In most cases this simple change is
            easy to apply as demonstrated below.
        </para>

        <programlisting language="php"><![CDATA[
/**
 * Before 1.10
 */
$feed = Zend_Feed_Reader::import('http://example.com/feed');
$authors = $feed->getAuthors();

/**
 * With 1.10
 */
$feed = Zend_Feed_Reader::import('http://example.com/feed');
$authors = $feed->getAuthors()->getValues();
]]></programlisting>
    </sect2>

    <sect2 id="migration.110.zend.file.transfer">
        <title>Zend_File_Transfer</title>

        <sect3 id="migration.110.zend.file.transfer.files">
            <title>Security change</title>

            <para>
                For security reasons <classname>Zend_File_Transfer</classname> does no longer store
                the original mimetype and filesize which is given from the requesting client into
                its internal storage. Instead the real values will be detected at initiation.
            </para>

            <para>
                Additionally the original values within <varname>$_FILES</varname> will be
                overridden within the real values at initiation. This makes also
                <varname>$_FILES</varname> secure.
            </para>

            <para>
                When you are in need of the original values you can either store them before
                initiating <classname>Zend_File_Transfer</classname> or use the
                <property>disableInfos</property> option at initiation. Note that this option is
                useless when its given after initiation.
            </para>
        </sect3>

        <sect3 id="migration.110.zend.file.transfer.count">
            <title>Count validation</title>

            <para>
                Before release 1.10 the <classname>MimeType</classname> validator used a wrong
                naming. For consistency the following constants have been changed:
            </para>

            <table id="migration.110.zend.file.transfer.count.table">
                <title>Changed Validation Messages</title>

                <tgroup cols="4">
                    <thead>
                        <row>
                            <entry>Old</entry>
                            <entry>New</entry>
                            <entry>Value</entry>
                        </row>
                    </thead>

                    <tbody>
                        <row>
                            <entry><constant>TOO_MUCH</constant></entry>
                            <entry><constant>TOO_MANY</constant></entry>

                            <entry>
                                Too many files, maximum '%max%' are allowed but '%count%' are given
                            </entry>
                        </row>

                        <row>
                            <entry><constant>TOO_LESS</constant></entry>
                            <entry><constant>TOO_FEW</constant></entry>

                            <entry>
                                Too few files, minimum '%min%' are expected but '%count%' are given
                            </entry>
                        </row>
                    </tbody>
                </tgroup>
            </table>

            <para>
                When you are translating these messages within your code then use the new constants.
                As benefit you don't need to translate the original string anymore to get a correct
                spelling.
            </para>
        </sect3>
    </sect2>

    <sect2 id="migration.110.zend.filter.html-entities">
        <title>Zend_Filter_HtmlEntities</title>

        <para>
            In order to default to a more secure character encoding,
            <classname>Zend_Filter_HtmlEntities</classname> now defaults to <acronym>UTF-8</acronym>
            instead of <acronym>ISO-8859-1</acronym>.
        </para>

        <para>
            Additionally, because the actual mechanism is dealing with character encodings and not
            character sets, two new methods have been added, <methodname>setEncoding()</methodname>
            and <methodname>getEncoding()</methodname>. The previous methods
            <methodname>setCharSet()</methodname> and <methodname>setCharSet()</methodname> are now
            deprecated and proxy to the new methods. Finally, instead of using the protected members
            directly within the <methodname>filter()</methodname> method, these members are
            retrieved by their explicit accessors. If you were extending the filter in the past,
            please check your code and unit tests to ensure everything still continues to work.
        </para>
    </sect2>

    <sect2 id="migration.110.zend.filter.strip-tags">
        <title>Zend_Filter_StripTags</title>

        <para>
            <classname>Zend_Filter_StripTags</classname> contains a flag,
            <varname>commentsAllowed</varname>, that, in previous versions, allowed you to
            optionally whitelist <acronym>HTML</acronym> comments in <acronym>HTML</acronym> text
            filtered by the class. However, this opens code enabling the flag to
            <acronym>XSS</acronym> attacks, particularly in Internet Explorer (which allows
            specifying conditional functionality via <acronym>HTML</acronym> comments). Starting
            in version 1.9.7 (and backported to versions 1.8.5 and 1.7.9), the
            <varname>commentsAllowed</varname> flag no longer has any meaning, and all
            <acronym>HTML</acronym> comments, including those containing other
            <acronym>HTML</acronym> tags or nested commments, will be stripped from the final output
            of the filter.
        </para>
    </sect2>

    <sect2 id="migration.110.zend.translate">
        <title>Zend_Translate</title>

        <sect3 id="migration.110.zend.translate.xliff">
            <title>Xliff adapter</title>

            <para>
                In past the Xliff adapter used the source string as message Id. According to the
                Xliff standard the trans-unit Id should be used. This behaviour was corrected with
                Zend Framework 1.10. Now the trans-unit Id is used as message Id per default.
            </para>

            <para>
                But you can still get the incorrect and old behaviour by setting the
                <property>useId</property> option to <constant>FALSE</constant>.
            </para>

            <programlisting language="php"><![CDATA[
$trans = new Zend_Translate(
    'xliff', '/path/to/source', $locale, array('useId' => false)
);
]]></programlisting>
        </sect3>
    </sect2>

    <sect2 id="migration.110.zend.validate">
        <title>Zend_Validate</title>

        <sect3 id="migration.110.zend.validate.selfwritten">
            <title>Self written validators</title>

            <para>
                When setting returning a error from within a self written validator you have to
                call the <methodname>_error()</methodname> method. Before Zend Framework 1.10 you
                were able to call this method without giving a parameter. It used then the first
                found message template.
            </para>

            <para>
                This behaviour is problematic when you have validators with more than one different
                message to be returned. Also when you extend an existing validator you can get
                unexpected results. This could lead to the problem that your user get not the
                message you expected.
            </para>

            <programlisting language="php"><![CDATA[
My_Validator extends Zend_Validate_Abstract
{
    public isValid($value)
    {
        ...
        $this->_error(); // unexpected results between different OS
        ...
    }
}
]]></programlisting>

            <para>
                To prevent this problem the <methodname>_error()</methodname> method is no longer
                allowed to be called without giving a parameter.
            </para>

            <programlisting language="php"><![CDATA[
My_Validator extends Zend_Validate_Abstract
{
    public isValid($value)
    {
        ...
        $this->_error(self::MY_ERROR); // defined error, no unexpected results
        ...
    }
}
]]></programlisting>
        </sect3>

        <sect3 id="migration.110.zend.validate.datevalidator">
            <title>Simplification in date validator</title>

            <para>
                Before Zend Framework 1.10 2 identical messages were thrown within the date
                validator. These were <constant>NOT_YYYY_MM_DD</constant> and
                <constant>FALSEFORMAT</constant>. As of Zend Framework 1.10 only the
                <constant>FALSEFORMAT</constant> message will be returned when the given date
                does not match the set format.
            </para>
        </sect3>

        <sect3 id="migration.110.zend.validate.barcodevalidator">
            <title>Fixes in Alpha, Alnum and Barcode validator</title>

            <para>
                Before Zend Framework 1.10 the messages within the 2 barcode adapters, the Alpha
                and the Alnum validator were identical. This introduced problems when using custom
                messages, translations or multiple instances of these validators.
            </para>

            <para>
                As with Zend Framework 1.10 the values of the constants were changed to
                be unique. When you used the constants as proposed in the manual there is
                no change for you. But when you used the content of the constants in your code
                then you will have to change them. The following table shows you the changed values:
            </para>

            <table id="migration.110.zend.validate.barcodevalidator.table">
                <title>Available Validation Messages</title>

                <tgroup cols="3">
                    <thead>
                        <row>
                            <entry>Validator</entry>
                            <entry>Constant</entry>
                            <entry>Value</entry>
                        </row>
                    </thead>

                    <tbody>
                        <row>
                            <entry><classname>Alnum</classname></entry>
                            <entry><constant>STRING_EMPTY</constant></entry>
                            <entry>alnumStringEmpty</entry>
                        </row>

                        <row>
                            <entry><classname>Alpha</classname></entry>
                            <entry><constant>STRING_EMPTY</constant></entry>
                            <entry>alphaStringEmpty</entry>
                        </row>

                        <row>
                            <entry><classname>Barcode_Ean13</classname></entry>
                            <entry><constant>INVALID</constant></entry>
                            <entry>ean13Invalid</entry>
                        </row>

                        <row>
                            <entry><classname>Barcode_Ean13</classname></entry>
                            <entry><constant>INVALID_LENGTH</constant></entry>
                            <entry>ean13InvalidLength</entry>
                        </row>

                        <row>
                            <entry><classname>Barcode_UpcA</classname></entry>
                            <entry><constant>INVALID</constant></entry>
                            <entry>upcaInvalid</entry>
                        </row>

                        <row>
                            <entry><classname>Barcode_UpcA</classname></entry>
                            <entry><constant>INVALID_LENGTH</constant></entry>
                            <entry>upcaInvalidLength</entry>
                        </row>

                        <row>
                            <entry><classname>Digits</classname></entry>
                            <entry><constant>STRING_EMPTY</constant></entry>
                            <entry>digitsStringEmpty</entry>
                        </row>
                    </tbody>
                </tgroup>
            </table>

        </sect3>
    </sect2>
</sect1>
<!--
vim:se ts=4 sw=4 et:
-->