Fixed ZF2-100: Use new naming convention for zend config options

[proofek]

No description provided.

[b-durand]

I disagree. Where come from this new naming convention?

[proofek]

See http://framework.zend.com/issues/browse/ZF2-100 for reference.

[b-durand]

When we say don't use underscore, we means $_foo not 'foobar' instead of 'foo_bar'. If we talk about ZendMonitor, you write this 'zend_monitor' and not 'zendmonitor'.

[sasezaki]

FYI.
current head's adapter has "headers key(ZF1 strtolower --> ZF2 ucrfirst-uchrfirst)" bug
http://framework.zend.com/issues/browse/ZF2-99

Socket Adapter's fix is
https://gist.github.com/1358388

If you run test tests/Zend/Http/Client/Adapter/*, "headers key" bug will block tests .

(I would pull-request for ZF2-99 fix later )

[weierophinney]

Closing -- intiilapa's comments hit the nail on the head. We're removing underscore prefixes for non-public variables and method names, but encouraging underscore separators for config options.

[proofek]

Well in that case library/Zend/Http/Client.php is broken, as it removes all underscores in config.

public function setConfig($config = array())
{
    if ($config instanceof Config) {
        $config = $config->toArray();

    } elseif (!is_array($config)) {
        throw new Exception\InvalidArgumentException('Config parameter is not valid');
    }

    /** Config Key Normalization */
    foreach ($config as $k => $v) {
        $this->config[str_replace(array('-', '_', ' ', '.'), '', strtolower($k))] = $v; // replace w/ normalized
    }

    // Pass configuration options to the adapter if it exists
    if ($this->adapter instanceof Client\Adapter) {
        $this->adapter->setConfig($config);
    }

    return $this;
}

[ralphschindler]

First, after the Options proposal is ratified, this will be a non-issue.

But secondly, more importantly, normalization allows for both 'keep_alive' and 'keepalive' to both be passed as a config option and all the subsequent classes the ability to read the value for that config value.

Also, I don't think this has anything to do with the @ZF2-99 that was reported, that seems like a case of keepalive being passed in, but also not waiting for the socket to finish what it is doing.

[proofek]

Well, if you gonna go for the support of both conventions then Proxy class needs to be changed to accomodate that. At the moment it's broken and will not proxy requests properly. Do you reckon it should check for both proxy_host and proxyhost options then?

[tcz]

I think an Adapter abstract should do the normalization in all cases. However, this is broken, so I think temporarily accepting the pull request above would be fine.

I came here after an hour debugging Mink+Goutte, which uses the entire ZF2 in order to function for some unclear reason.