After creating a metadata factory and metadata drivers, we now need a way to cache the metadata, since collecting them at each request is way too inefficient. Luckily, the Metadata library provides us with a FileCache (though it may be any kind of a cache, as long as it implements the same CacheInterface). First, we make the FileCache available as a service, and add a call to setCache on the metadata factory:

<parameter key="matthias_annotation.metadata.cache_class">Metadata\Cache\FileCache</parameter>

<!-- ... -->

<service id="matthias_annotation.metadata.cache" class="%matthias_annotation.metadata.cache_class%" public="false">
    <argument /><!-- the cache directory (to be set later) -->
</service>

<service id="matthias_annotation.metadata_factory" class="%matthias_annotation.metadata_factory.class%" public="false">
    <argument type="service" id="matthias_annotation.driver_chain" />
    <!-- call setCache with the new cache service: -->
    <call method="setCache">
        <argument type="service" id="matthias_annotation.metadata.cache" />
    </call>
</service>

This provides the metadata factory with the file cache.

In my previous post, I wrote about annotations and how to gather them using a metadata factory and an annotation driver. I also mentioned a future post in which I explain other ways of collecting metadata, namely by locating and parsing files in several different formats (distinguished by their extension). This is the post. In the case of our @DefaultValue annotation, I’m thinking of a Yaml flavoured alternative. This allows us to store all default values as an array of property “name” - “value” pairs, like