a programmer's best friend :: Entries from December 2004

I released new versions of two of my PEAR packages: XML_Parser 1.2.2 and XML_Serializer 0.14.0. While XML_Parser 1.2.2 only fixes two bugs, XML_Serializer 0.14.0 provides two cool new features:

1. By splitting the SAX-handlers from the XML_Parser class (which is possible since XML_Parser 1.2.0) you may now set the source and the target encoding of the XML documents that you want to unserialize. This enables you to read UTF-8 encoded documents and convert them to ISO-8859-1 on-the-fly.
2. I added a new option "encodeFunction" to the XML_Serializer class, which lets you specify a PHP function or method that will be called for all character data and attributes before they are written to the XML document. This allows you to encode your data in UTF-8 before creating an XML-document. As XML_Unserializer provides a matching option to decode the data, you could also use this to transparently encrypt XML and decrypt XML documents

I added these two new features, as eBay requires applications to work with UTF-8 encoded XML documents and thus Services_Ebay needs to read and write UTF-8 encoded XML-streams. But the more I think about these features the more situations come to my mind where those options could be timesavers.

This week, two of my articles has been published. It started with issue 1.05 of the German "Java Magazin" for which I did a short introduction of to XUL. XUL is an XML language to model user interfaces for Mozilla, I did a PEAR class that allows you to create XUL applications with PHP.

Today I received the latest issue of the International PHP Magazine, which includes my article on XML_Serializer, I package I maintain in PEAR that is able to serialize almost any PHP structure to an XML document and vice versa.

I just released a new version of Services_Ebay, which fixes a small bug in the Item Model, allows you to change the language of the error messages and allows you to specify international shipping options for your items.
But the most important change is, that Adam Trachtenberg now has CVS karma to commit to the CVS repository. He already commited his first patch, hope there will be some more in the near future, when he started working with the package. So welcome aboard, Adam!

I just proposed a new package for inclusion in PEAR: HTTP_Cache. This package helps you creating ConditionalGet requests in your pages. This helps you keep the traffic low on your sites as it only sends the request body with the first request and a 304 (Not Modifed) header on subsequent requests. This is especially useful, if a page that is requested quite often is not changed a lot, but still is generated by PHP, so the browser will not cache it by default.
If you would like to implement this in your site, HTTP_Cache does all the work for you and even is able to work with output buffering. If using this feature, you only need to add two lines of code to the scripts you want to cache:
<?php
require_once 'HTTP/Cache.php';
$cache = &new HTTP_Cache(array('auto' => true));

echo "You now may send any data to the browser";
?>
HTTP_Cache will generate a unique id for the current browser and send the following headers:
Cache-Control: must-revalidate
Etag: e33b111743ff17c7829cd6bdad6815cb

On the next request the client will resend the ETag back to the server and HTTP_Cache will compare the client's ETag with the one that will be generated on the server. If both tags match, the data will not be send to the client, but a 304 header will be issued:
HTTP/1.x 304 Not Modified

The only drawback is that HTTP_Cache will need to create an MD5-sum of the content, which could slow your application down in some cases. But using the cache will speed up your site as your webserver does not have to send as much data as without the HTTP_Cache.
Furthermore you may supply the unique id so HTTP_Cache does not need to create use the md5() function. This is especially helpful, when you are already using a serverside cache system and already generated a unique cache key for the current page. In this case you may also check whether the browsercache is valid so you do not need to load the cache file from disk:
<?php
require_once 'HTTP/Cache.php';

$cache = &new HTTP_Cache();

// create an etag
$etag = 'hfhfhjfjddhfhjdshdddkjd33';
$cache->setEtag($etag);

// The browser cache is not valid
if (!$cache->isValid()) {
// create your content
$html = 'Do some expensive HTM creation....';

// pass it to the cache
$cache->setBody($html);
}

// send header or data
$result = $cache->send();
?>
If you are interested in this package, take a look at the package proposal or download the first version from out site.

While shopping for christmas presents on Saturday, I found Christian's and Tobias' new book PHP5 Kompendium standing on the shelves and until then I hadn't realized, that this already had been published. The German book deals with all you need to know about PHP5 and is the perfect christmas present for everybody who either wants to learn PHP from scratch or is looking for some advanced techniques.
While skipping through Christian's blog, I discovered that the book had been released nearly two weeks ago, I guess it's time that Christian's blog is included in Planet PHP.

Congratulations to Christian and Tobi.

In the latest issue of the PHP Barnstormer Aaron wrote a nice summary of Greg's tutorial on how to use those strange phpt-files and how to write your own tests. This currently is the best source of information on phpt files you can get (at least I could not find better information).
Together with the new PEAR_RunTest class, that Greg wrote, it will finally be a lot easier using these standard tests for your PEAR packages.
This reminded me, that a lot of my PEAR packages still are in need of some good unit tests, guess I'll give phpt a try.