I'm using the DebugKit component in my project but I want to turn it on only on the staging server and have it not load when running from the production server.
I know I can turn it off with the debug config value but I want to keep that value at 1 for both servers.
I tried conditionally defining a contstant 'DEBUG_KIT' in bootstrap.php as either the component name (ie. 'DebugKit.Toolbar') or null. Then using that constant in the var $component definition at the top of the app_controller file. Well, Cake doesn't like having a null in the component array and barfs. Doesn't like an empty string either.
I feel like I'm missing something but can't quite see the forest for the trees. Thanks in advance!
I think the fundamental purpose of DebugKit is tied to being in debug mode, so I can understand that the tools don't provide the capacity to be disabled without also disabling debug mode.
That said, if you absolutely must do this, I think your best bet is to directly modify app/plugins/debugkit/controllers/components/toolbar.php, supplementing the existing debug-mode check in ToolbarComponent::initialize with a check against your constant.
(For what it's worth, I think you'd be better off turning debug-mode off on your production server, and using errors/warnings logged in /app/tmp/logs/error.log to identify issues that have slipped through your testing.)
I do something similar in my apps: I would use the __construct method to detect the presence DEBUG_KIT and add it to the $components array. This function gets called before the $components array is processed, so you can add/remove components transparently.
In your app_controller
function __construct(){
if(DEBUG_KIT){
$this->components[] = 'DebugKit.Toolbar'
}
parent::__construct();
}
If you have a _construct function in any or your individual controllers, remember to include parent::_construct(); otherwise you'll 'break the chain'.
Hope this helps
First, thanks to Adam Giles for a great answer. I hadn't thought to look at the __construct() callback. That just may be a better way than I found. And to Daniel Wright, point made sir. I'll probably change my production server to 0 debug soon and start watching the error logs.
I found my own answer shortly after posting this question. The DebugKit has an 'autoRun' parameter that will turn it on and off. So, I first set a global constant in bootstrap.php like this:
define( 'IS_DEV', ($_SERVER['SERVER_NAME'] == 'staging.example.com') );
Then in app_controller.php, I use it to set the 'autoRun' parameter in the $components statement.
var $components = array( 'DebugKit.Toolbar'=>array('autoRun'=>IS_DEV) );
This seems to work pretty well so far.
Related
I was following documentaion till the end, tested one html template with smarty and then cut it. Then I found out that controllers do not work as expected – whatever name I create in myapp/conrollers, 'hello.php' for example, that contains class described in docs, i. e.
class Hello_Controller extends TinyMVC_Controller
{
function index()
{
echo "Hello World.";
}
function time()
{
echo "The time is now.";
}
}
I can’t show it. So the name of the file is a prefix for the controller class name, all seems to be ok here, but going to /index.php/hello returns what is in 'default.php'. I’ve even tried to change default controller to 'hello' in myapp/configs/application.php by setting $config['default_controller'], but the framework behaves like if it’s always work with the 'default.php'. There is no errors on screen or in logs (I checked twice every option in configs of my web server and interpreter), I totally don’t know what to do with that goddamn piece of crap, I can’t even write on its forum because waiting for ‘administration approval’ for several days.
I had to dig inside of the framework to find an answer. And it is when it checks for a controller file it uses file_exists() which do not respect include path. Googling ‘TinyMVC+file_exists’ gave me link to that topic, where is written that they had fixed it in SVN version.
I'm making a web application in Zend Framework. I've reached the stage of cleaning up. As things often go, I have a couple of messy view scripts that have become utterly unreadable (tons of (v)sprintf's and loops).
There's one view that's an absolute nightmare... (no/inaccurate comments, shorthand... all things considered to be mortal sins). Just an example:
$rows[$c] .= '<div>'.sprintf('<select id="%s" name="%1$s">',$t.'['.$ref->getCode().']').str_replace('>'.$ref->getCValue().'<',' selected="selected">'.$ref->getCValue().'<','<option>'.implode('</option><option>',$this->vals['P']).'</option>').'</select></div>';
In this particular case, I have an array of models that looks like this:
$arr = array('FOO'=> $Mylib_Model_Person,'BAR'=> $Mylib_Model_Person2);//1~50 mdls
I would like to use a partial loop, but there's a problem:
$this->partialLoop('controller/_myPartial.phtml',array('model'=>$arr));
//in the partial loop:
Zend_Debug::dump($this->m);
I see all my models correctly, but their keys have all been turned into properties.
$this->FOO->someMethod();//works fine
Bur I want it to be:
<span><?php echo $key; ?></span><span><?php echo $model->someMethod(); ?></span>
I've also tried using $this->partialLoop()->setObjectKey('Mylib_Model_Person');, but that didn't seem to make any difference, other then confuse me.
The only solutions I see is either array_map, but that would defeat the point (I'm trying to end up with a clean view script); or rewrite a part of my service layer, to return the data ready structured, and keep the array_map there.
I can't help thinking that what I want to do, essentially use a partialLoop as an array_map callback, is possible. If it isn't, is there an alternative? Any thoughts?
I've tried get_object_properties($this), and iterate through the object properties, to no avail, the loop simply doesn't get executed(?!)
As it turns out $this->partialLoop()->setObjectKey('Mylib_Model_Person'); should have been $this->partialLoop()->setObjectKey('model');. If I do change this, and start the partial loop by dumping $this->model, I see my model. However:
echo $this->model->someMethod(); //throws error: method on non-object
Zend_Debug::dump(get_class_methods($this->model));//shows all methods, including someMethod()
And to add insult to injury, tears and confusion. The model implements the toArray-thing, so I tried:
echo $this->model['someData'];//Error: cannot use object of type Mylib_Model_Person as array!!
So, it's not an object when I try to use methods, it's an object when trying to access data as an array, and when using the magic getter method ($this->model->some_Data) it doesn't do anything. No errors, but no output either. The view is rendered as is.
I'm thinking I ran into a bug. I'll rapport it. Consider this:
$methods = get_class_methods($this->model);
while($m = array_shift($methods))
{
if (substr($m,0,3) === 'get')
{
Zend_Debug::dump($m);//e.g getName
Zend_Debug::dump($this->model->{$m}());//'Foobar'
$m = 'someMethod';//copy-paste, so typo's aren't to blame
Zend_Debug::dump($this->model->{$m}());//'the data I was after'
}
}
So that works, but the, if I try:
$this->model->{'someMethod'}();//Error again
//or even:
$m = 'someMethod';
echo $this->model->{$m}();//Error...
That can't be right
I found out what the problem was. Our development server used to be set up better in terms of error reporting:
I assumed E_ALL | E_STRICT, but I had a look only to find it changed to a cruddy E_COMPILE_ERROR | E_ERROR | E_CORE_ERROR. Seeing as some values in the array can be false rather then an object, at some point in the partialLoop script, a notice should have been raised - with the correct ini settings, of course.
That was the cause of the unexpected behaviour; that, and a silly bug or two.
When I'm trying to execute a functional unittest of a module within my Yii code, I keep receiving the following error:
CException: CHttpRequest is unable to determine the request URI.
At first, I though it was because it couldn't find the module. However, If I change the url to a wrong one, I get a correct error,s tating it couldn't find the view.
This is how my testing code looks like
public function testViewControllerModule()
{
ob_start();
Yii::app()->runController('module/controller/view');
}
Any ideas on what I might be missing?
bool.devs answer works so far.
This blog post explains the origin of the exception pretty well:
http://mattmccormick.ca/2012/09/14/unit-testing-url-routes-in-yii-framework/
In my case, I generalized the solution and have set the following variables in /www/protected/tests/bootstrap.php:
...
$_SERVER['SCRIPT_FILENAME'] = 'index-test.php';
$_SERVER['SCRIPT_NAME'] = '/index-test.php';
$_SERVER['REQUEST_URI'] = 'index-test.php';
Yii::createWebApplication($config);
Consider using 'index-test.php' instead of 'index.php' because it contains the config 'test.php' which is responsible for fixtures and maybe other test relevated configurations.
If someone has better suggestions feel free to comment :)
Kind regards
I think it's because you haven't set any server variables, i.e $_SERVER and you might be doing something like this in your controller:
Yii::app()->request ....
So before you run your test, make sure you use a fixture for the server variables also. I think this should suffice for now:
$_SERVER=array(
'REQUEST_URI'=>'index.php', // the other fields should follow
);
However to run functional tests i would recommend using SeleniumRC, you won't have to do these workarounds then, and can simulate user clicks also, i think.
Read the initial guide to Functional Testing , read the selenium rc phpunit guide, and also the CWebTestCase documentation.
Notes: You might still have to use fixtures for some variables, and i don't have much experience in testing(which is bad), so i'm not very sure if i am completely correct about selenium.
We have used Zend_Log, which is configured in application.ini differently for different circumstances. We initialize it/get it in the bootstrap and store it in the registry:
$r = $this->getPluginResource('log');
$logger = $r->getLog();
But we've subclassed Zend_Log (say, Our_Log) to add customized features, and want to get it the same way. So then we have to make a new Resource Plugin. That seems quite easy - just copy Application/Resource/Log.php, rename the file to Ourlog.php, rename the class to class Zend_Application_Resource_Ourlog. For now, let's not worry about "Our_Log", the class -- just use the new Resource Plugin to get a Zend_Log, to reduce the variables.
So then, our new code in the bootstrap is:
$r = $this->getPluginResource('ourlog');
$logger = $r->getLog();
but of course this doesn't work, error applying method to non-object "r". According to the documentation,
"As long as you register the prefix path for this resource plugin, you
can then use it in your application."
but how do you register a prefix path? That would have been helpful. But that shouldn't matter, I used the same prefix path as the default, and I know the file is being read because I "require" it.
Anyway, any guidance on what simple step I'm missing would be greatly appreciated.
Thanks for the pointers -- so close, so close (I think). I thought I was getting it...
Okay, so I renamed the class Xyz_Resource_Xyzlog, I put it in library/Xyz/Resource/Xyzlog.php
Then, because I don't love ini files, in the bootstrap I put:
$loader=$this->getPluginLoader();
$loader->addPrefixPath('Xyz_Resource','Xyz/Resource/');
$r = $this->getPluginResource('xyzlog');
if (!is_object($r)) die ('Not an object!!');
Sudden Death. So, okay, do the ini:
pluginPaths.Xyz_Resource='Xyz/Resource/'
The same. No help. I believed that the basepaths of the plugin paths would include the PHP "include" paths. Am I mistaken in that? Any other ideas? I'll happily write up what finally works for me to help some other poor soul in this circumstance. Something to do with Name Spaces, maybe?
Plugin classes are resolved using the plugin loader, which works slightly differently to the autoloader; so just requiring the class in doesn't help you here. Instead, add this to your application.ini:
pluginPaths.Application_Resource = "Application/Resource"
you should then be able to use the class as normal. Since your path above will be checked before the default Zend one, you can also name your class 'Log' and still extend the Logger resource to override the standard functionality.
I am learning how to use codeIgniter as my php framework. I am reading through the documentation and watching the intro video and just generally following along with the first tutorial, but it's not working for me.
I have created a controller called "test.php" and a view called "test_view". The controller for this class is exactly like "welcome.php" and the view file just has some static html. However, when I go to index.php/test I get a 404 error.
I have also tried manipulating the original welcome files so that instead of calling a view it just echos "testing", yet I still see the original welcome message! I've tried clearing my browsing cash and refreshing, but to no avail.
Any suggestions? Thanks.
Edit: Here's the code for controllers/test.php
<?php
class Test extends Controller {
//Just trying to get it to echo test
public function index()
{
echo "test";
//$this->load->view('test_view');
}
}
?>
Try looking at this page in the documentation - this might solve your problem.
This basically means you should try typing index.php?/test/ instead (notice the question-mark).
First of all, check the above link. Might be useful.
If not, then...
Try changing the default controller in the config file ('routes.php') to something else (probably, to 'test'), then try loading index.php. Just to test whether the whole system works (or not).
Check whether mod_rewrite is loaded (in your server .conf-file, if you're using Apache).
Try using the latest build of the framework. AFAIK, the name of the controller class for now is "CI_Controller".
Finally, try removing the word 'public' before the declaration of the function. AFAIR, CI enable you to make private functions in controllers just by using prefix (which is set in the config file) at the beginning of the name of the function (thus making all the other functions public).
But most certainly the problem is with the mod_rewrite. If not, try debugging with die('Page found'); instead of echo - this will allow you to track possible redirects on the page.