new_way.php

<?php
use \eZ\Publish\API\Repository\Values\Content\Query;

$locationService = $this->getRepository()->getLocationService();
$searchService = $this->getRepository()->getSearchService();
$urlAliasService = $this->getRepository()->getUrlAliasService();
$typeService = $this->getRepository()->getContentTypeService();

$identifiers = array( 'folder', 'page', 'contact' );
$ids = array();
foreach ( $identifiers as $identifier )
{
    $ids[] = $typeService->loadByIdentifier( $identifier )->id;
}

$root = $locationService->loadLocation( 2 );
$query = new Query();
$query->criterion = new Criterion\LogicalAND(
    array(
        new Criterion\Subtree( $root->pathString ),
        new Criterion\ContentTypeId( $ids ),
        new Criterion\Visibility( Criterion\Visibility::VISIBLE )
    )
);
$query->sortClauses = array(
    new SortClause\LocationPriority( Query::SORT_DESC )
);
$results = $searchService->findContent( $query );
$items = array( );
foreach ( $results->searchHits as $hit )
{
    $location = $locationService->loadLocation(
        $hit->valueObject->contentInfo->mainLocationId
    );
    $locationUrl = $urlAliasService->reverseLookup( $location )->path;
    $items[$locationUrl] = $location;
}

foreach ( $items as $url => $location )
{
    echo "url: {$url} name: {$location->contentInfo->name}\n";
}

old_way.php

<?php
$items = eZContentObject::subTreeByNodeID(
    array(
        'ClassFilterType' => 'include',
        'ClassFilterArray' => array( 'folder', 'page', 'contact' ),
        'SortBy' => array( array( 'priority', false ) )
    ),
    2
);

foreach ( $items as $node )
{
    echo "url: {$node->attribute( 'url_alias' )} name: {$node->attribute( 'name' )}\n";
}

Looking at this I think there is a place for a legacy data access layer on top of the new public API. It would decrease the verbosity of the code required to do the retrieval & aid in converting to the new platform.

e.g. Mimic the functionality of eZContentObject::subTreeByNodeID but use the new Public API to do the work.

Actually... There is much more code. But it's easier to extend and improves testability thanks to dependency injection. Here, you are building everything in the same script. But it's possible to have a set of defined queries:

$results = $searchService->findContent(  new MyArticlesQuery );

or:

$results = $searchService->findContent(  new MyConfigurableQuery('articles') );

or:

$results = $searchService->findContent(  new MockObjectForTestPurpose );

This was not possible with eZContentObject::subTreeByNodeID()

So... I'm still confident in this new API ;)