Posted on

Using Airtable through its API programmatically, as (almost) remote database

I recently talked a lot about the importance of collaborative and smarter documentation that will improve your personal and professional workflow. Certainly, it will be different than other competitors in an interesting use case I found myself in one of the hobby projects that I used Airtable as a remote database tool all of the sudden.

Airtable is a very nice mobile-friendly document management in a “spreadsheet” style base. You can create your data structure in any data table model. You can create different views for your data (in calendar view, or filtered table view, or kanban view…).

img_0201.jpg

What makes Airtable special for me is its API. Their API is so easy to get started and access, because you get your data-specific API documentation after you login. It shows your data in API example right and there in the dynamic documentation.

img_0200.png

Airtable API essentially makes Airtable that can be used as a remote database for small or personal projects. Managing your data belongs to Airtable’s clean and nice web or mobile interface, and you implement your data in any format you like on any platform.

If you are needing read-only access, implementing Airtable API can be a matter of minutes since the documentation gives access to your data very quickly. You only need to convert the curl request to your favorite platform’s HTTP request. If you are needing a javascript version, it also produces NodeJS example code that you can drop in and start using your data.

img_0199.png

Write access is also not very different than read-only. Your data model will be well be documented in the dynamic API documentation for your table. You only need to start constructing your API requests and make the call…

If you haven’t created an Airtable account and played with it, definitely do so: https://airtable.com/ and check out their auto-generated documentation here: https://airtable.com/api (after you login to your account).

Posted on

Profiling and optimization on Facebook PHP SDK

If you’re developing PHP based Facebook application, you might want to use (or already using) Facebook integration little more than just authentication and identification of your user. Even if you have the simplest Facebook app and using PHP SDK, you probably have regularly done API calls.

You write your app and you start to see performance issues. You start to optimize your database interactions, PHP code optimization, after you done with your application optimization if you still have performance problems it’s possibly from Facebook calls. Since you use SDK, you might not know what’s happening with Facebook communication. So you want to do profiling between your app and Facebook API servers.

You can add a basic timing profiling to your API calls to see how many calls you do, what kind of calls they are and how long they take to run.

Let’s dive in SDK, modify it a bit and start getting profiling information. Here is the actual method you need to modify in base_facebook.php file:

public function api(/* polymorphic */) {
	$args = func_get_args();
	if (is_array($args[0])) {
		return $this->_restserver($args[0]);
	} else {
		return call_user_func_array(array($this, '_graph'), $args);
	}
}

and we’re modifying it like this:

$facebookApiCalls = array();
public function api( /* polymorphic */)
{
	$args = func_get_args();

	$time_start = microtime(true);

	if (is_array($args[0])) {
		$result = $this->_restserver($args[0]);
	} else {
		$result = call_user_func_array(array($this, '_graph'), $args);
	}

	$time_end = microtime(true);
	$time_elapsed = $time_end - $time_start;
	$time_elapsed *= 1000; //convert to millisecs

	if (isset($GLOBALS['facebookApiCalls'])) $GLOBALS['facebookApiCalls'][] = array(
		'duration' => $time_elapsed,
		'args' => $args,
	);
	return $result;
}

It simply appends a global array named “facebookApiCalls” and adds the API call details as “args” and timing as “duration”. So at the end of your page logic code, you can print this information after sorting it by duration and you can also filter to show only slow ones (for instance, the ones took over 200 milliseconds).

After this profiling you can start to identify the slow calls, also if you do same calls multiple times because of recursing, recalls etc…, you can see and optimize, combine them.

This optimization is not only a performance tweak for the user, also it will decrease the number of calls made between your server and Facebook servers.