302 Found

iPhone DTMF woes

2009-10-01T15:33:00.003+03:00

Unfortunately, every other day is equal to another iPhone horror story for me. Developing an Asterisk IVR (Interactive Voice Response) system means interpreting DTMF tones from a wide range of devices. Now, if one of them happens to be an iPhone... then all hell breaks loose.

While testing, we weren't exactly sure why the keypad input from one of our phones was always garbled. After a few minutes of testing and testing, we figured it must be a phone problem. And it was. The phone in question is (of course) an Apple iPhone 3G. *My* iPhone 3G.

Digging through the apple support forums, this is what came up: the iPhone only knows how to send "long" DTMF tones. Because phone line quality is lousy by default, line noise and interruptions cause these long tones to sometimes (almost always) get broken into two or more "short" tones. Hence, the keypad input nightmare begins.

Empirically we discovered that turning off the 3G service on the iPhone alleviates the problem, but it's not guaranteed to work, and it's not documented anywhere.

Did I mention I have a love-hate relationship with my iPhone?

Fogbugz > Redmine

2009-09-30T15:44:00.002+03:00

Fogbugz is orders of magnitude better than Redmine. Period. Also a lot more expensive. But it's worth every penny.

Just needed to say that.

Oh, and if you're a startup or a student, you should try their Fogbugz Student and Startup Edition (DUH).

Redirect Non-www Domain to www for Better SEO

2009-03-30T23:46:00.000+03:00

Search engines consider http://domain.com and http://www.domain.com different websites. As a result, you are effectively splitting the potential benefit of valuable link popularity, as well as risking to be penalized by Google for duplicate content.

Using a permanent redirect you can effectively consolidate all of your link popularity to a single URL.
This Search Engine Optimization will serve to increase your website's chances of obtaining and maintaining top rankings.

First of all, you need to decide which version you intend to keep: the "www" or "non-www" URL for your website. Depending on which version you intend to keep, you need to use the following code in your Apache configuration files:

Redirect domain.com to www.domain.com

RewriteEngine On RewriteBase / RewriteCond %{HTTP_HOST} !^www.domain.com$ [NC] RewriteRule ^(.*)$ http://www.domain.com/$1 [L,R=301]
Redirect www.domain.com to domain.com

RewriteEngine On RewriteBase / RewriteCond %{HTTP_HOST} !^domain.com$ [NC] RewriteRule ^(.*)$ http://domain.com/$1 [L,R=301]

Some hosting providers are more permissive than others. If you have access to your httpd.conf file, you can insert the code there. If not, you can always use the .htaccess file in your site's document root. See the post
How to Disable Directory Listing/Browsing in Apache Web Server for more details.

Also, you need to ensure that your hosting provider has the Apache Rewrite Module turned on; this is a definite requirement for this fix to work. You can enable the Rewrite Module using the following line in the configuration files:

LoadModule rewrite_module modules/mod_rewrite.so

How to Disable Directory Listing/Browsing in Apache Web Server

2009-03-21T21:43:00.004+02:00

Apache's Web Server, HTTPD, is arguably the most popular choice for a website, with more than 66% of the busiest websites using it (according to Netcraft's March 2009 Web Server Survey). By default, it comes configured with the Directory Listing option enabled. This option allows users to see all the files and sub-directories under a particular directory when index files set by DirectoryIndex (e.g., index.html, index.php etc.) are not found in that directory.
This is not advisable on a production website, because it may expose sensitive information to prying eyes. Moreover, if you do not define entries in robots.txt properly, search engines can reach and index these directories. Therefore, it is necesary to make these directories invisible in order to protect your data from being compromised.

How can we disable the Directory Listing in Apache Server?

There are two ways to achieve this:

You can do this with an Options directive inside a <Directory> tag. Either remove the Indexes option from the line that begins with Options if it’s there, or change Indexes to -Indexes.
```
<Directory "/www">
Options Indexes FollowSymLinks
</Directory>
```
The only disadvantage with this method is that you need to have access the Apache configuration file, httpd.conf.
Since most shared hosting companies don't allow this, you might not be able to use the previous method. You can, however, obtain the same effect as method 1 by using the .htaccess file: open the .htaccess file (create a new one if it’s not there) under the target directory, look for Options Indexes. If Options Indexes exists, modify it to Options -Indexes, or else add Options -Indexes as a new line.

This method requires the AllowOverride directive be set to "Options" or "All" for the target directory. When this directive is set to "None", then .htaccess files are completely ignored. In this case, the server will not even attempt to read .htaccess files in the filesystem.

Remember to restart your Apache Web Server, and the Directory Listing feature should be disabled by now.

Zend Studio for Eclipse 6.1.1 for Mac OS X - problems & solutions

2009-03-15T22:30:00.009+02:00

I recently reinstalled Leopard on my laptop, not because of an OS problem, but because my hard-drive quit on me. This was the perfect opportunity for me to try out version 6.1.1 of Zend's IDE. This, and the fact that version 6.1.0 that I used before was not available anymore.
To my displeasure, I was confronted with a handful of problems with this version of the IDE.
After enthusiastically downloading and installing the ~300 MB package (shame on you, Zend, for not providing links for the older versions of your products), I was greeted by a not-so-familiar face. All of the fonts looked really messed up, large and clunky. This was annoying for a number of reasons. First off, the larger fonts used up a large amount of real-estate on my 13" display. More importantly, it was plain ugly :(
I took to the Preferences dialog to try and modify the fonts, but to no avail: there were parts of the interface that stubbornly refused to comply with my settings. Thinking that maybe there was a problem with my Java installation, I downloaded and installed Eclipse, and noticed that it was NOT affected by the font problem. After a couple of hours of comparing Eclipse and ZSE configuration files, I discovered the cause. The line

-Dorg.eclipse.swt.internal.carbon.smallFonts

was missing from /Applications/Zend/Zend Studio for Eclipse - 6.1.1/ZendStudio.app:MacOS/ZendStudio.ini

The same comparison helped me figure out how to change the ugly low resolution icon that appeared for ZSE when pressing switching apps by using ⌘+Tab. Adding the line

-Xdock:icon=../Resources/mac.icns

to the same file as above made OS X use the high res version of the icon.

The next problem reared it's ugly head when I tried updating ZSE. Whatever options I'd chose when trying to install new features or update, the following message would appear in the Search Results dialog:

RSE Feature Patch (3.0.2.v20090119-1307w311_122_) requires feature "org.eclipse.rse (3.0.2.v200812041720-7H3788qfeqHMeUOEQ2IKjVLIiQ8)".

Fortunately, the solution to this problem was easily discovered on Zend's site in a Knowledge Base article.

In spite of all these minor problems, ZSE still is, in my opinion, the best PHP IDE out there. However, Zend should take a little more care when packaging new releases of their products.

My favorite project gets a new face

2009-02-12T16:14:00.003+02:00

After almost three months of non-stop work and a lot of interesting technical challenges, trafic.ro got a major redesign. But it's not all about the looks. The interface is cleaner and a lot of functionality has been updated to provide better usability.
And, judging from the initial feedback from our users, the new AJAX user interface for ranking system is a BIG hit!

Launched in September 2000, trafic.ro is the audience monitoring and website ranking tool used by most Romanian websites.

Check it out for yourself!

Back!

2009-01-17T04:06:00.001+02:00

It's been a long time since I've been here!
Good news is that starting January 15th I'm a WAA Professional member.

Selective Expandable Post Summaries

2008-06-24T11:29:00.004+03:00

I added Selective Expandable Post Summaries to my Blogger template today (the "Read more..." link you see on some larger posts in my list). This was a fairly simple exercise, thanks to an excellent article in Blogger Magz. The trick is to use Blogger's conditional tags, and a little javascript magic.

Cross-browser HTML/CSS and browser snapshots

2008-06-19T14:56:00.009+03:00

Even in this day of mature standards and strict implementation guidelines, having your web page look the same across the wide variety of browsers and operating systems available is still proving to be a major challenge for web designers and programmers alike.

In order to overcome the differences between browsers, webmasters usually have virtual machines with one or more different operating systems installed, each loaded with specific browsers. Dedicated design companies often keep "browser farms": racks of physical machines whose sole purpose is to provide different browsing environments. Each machine is accessible using VNC or Remote Desktop, and as such, every developer in the company can use the "browser farm" to check his designs. Such solutions demand significant hardware resources, not to mention the amount of time required to set up and manage the virtual or physical machines used for cross-browser checks.

A different approach to the problem is using a dedicated web service for capturing screenshots of websites of your choice.

http://www.browsershots.org/ is a free open-source project, invented and implemented by Johann C. Rocholl. Public beta test for the site started in February 2005 and the first public release was in March 2005.

The central idea behind browsershots.org the same that drives other community projects, like seti@home or folding@home. From the website's FAQ:

This is the one feature that sets this project apart from similar online services. The screenshots are made on distributed computers that are run by volunteers. You can see a list of active screenshot factories. Most of these are normal desktop machines with domestic broadband internet. Distributed community processing ensures O/S and browser diversity, but it makes quality assurance much harder.

Although the service is free, there is also a priority service, which you can have access to for 10€ ($15) per month. This means that your URLs don't have to wait in the job queue, they will be processed before all other requests. You will usually get 10 to 20 screenshots within 1 to 3 minutes.

http://www.browsercam.com/ is another browser snapshot website. They offer their bank of testing machines remotely to test your website.

BrowserCam's Screen Capture Service lets you submit single or multiple URL's, choose the browsers and operating systems you want to see, and screen captures of your webpage are loaded in the different browsers and operating systems you selected.
You may take an unlimited number of our traditional single-frame screen captures during your subscription period. Our screen capture service also offers Full Page Captures, which will take a full length screen capture up to 8 pages long.

One interesting feature they offer as an extra is the ability to check the way web pages display on mobile devices. Only Blackberry, Windows Mobile 5.0, Opera 8.65 at the moment, but they have plans for Windows Mobile 6.0 as well as a few different platforms. If you're designing with mobile users in mind, BrowserCam is definitely worth a try.

Their services cost $39.95 per month for standard web capture, and $59.95 per month if you also require mobile web browser snapshots. They also offer a free trial, which includes 24 hours of free usage of the screen capture service or a maximum of 200 screen captures.

Yay for me!

2008-02-27T19:34:00.001+02:00

It's official! I'm a Zend Certified Engineer!

REAL programmers use butterflies

2008-02-12T16:37:00.000+02:00

nano? real programmers use emacs
Hey, REAL programmers use vim
Well, REAL programmers use ed
No, real programmers use cat
REAL programmers use a magnetized needle and a steady hand
Excuse me, but REAL programmers use butterflies

Real Programmers via xkcd - A webcomic of romance, sarcasm, math, and language.

Oracle Zend_Db performance improvement

2008-02-05T09:19:00.001+02:00

The fun part about Zend_Db is that it gives you a consistent interface to the database across a wide range of RDBMS and applications. The not-so-fun part is that this power comes at a price, and that price is performance.

Suppose you have this piece of php code:


$query = "
        SELECT * 
        FROM source_table
        WHERE source_table.id = :id
";
for($i=0;$i<1000;$i++)
{
    $bind = array( ':id' => $i );
    $results = $db->fetchAll( $query, $bind );

    //do smth with the results
}

What about it? you say... Well, it's nice, clean, and also inefficient. Let's take a look at Zend_Db_Adapter_Abstract::query() (which is called from the fetchAll() method)


     public function query($sql, $bind = array())
    {
        // connect to the database if needed
        $this->_connect();

        // is the $sql a Zend_Db_Select object?
        if ($sql instanceof Zend_Db_Select) {
            $sql = $sql->__toString();
        }

        // make sure $bind to an array;
        // don't use (array) typecasting because
        // because $bind may be a Zend_Db_Expr object
        if (!is_array($bind)) {
            $bind = array($bind);
        }

        // prepare and execute the statement with profiling
        $stmt = $this->prepare($sql);
        $stmt->execute($bind);

        // return the results embedded in the prepared statement object
        $stmt->setFetchMode($this->_fetchMode);
        return $stmt;
    }

As you can see, our query is prepared every time we call Zend_Db_Adapter_Oracle::query(). What's the point of doing that, since we're doing the same query 1000s of times?
Here's the improved version:


$query = "
        SELECT * 
        FROM source_table
        WHERE source_table.id = :id
";
$stmt = $db->prepare($query);

for($i=0;$i<1000;$i++)
{
    $bind = array( ':id' => $i );
    $stmt->execute( $bind );
    $results = $stmt->fetchAll( $db->getFetchMode() );

    //do smth with the results
}

There you go! Instant performance improvement, with minimal code modifications.

Oracle Text Indexing

2008-02-05T02:18:00.001+02:00

Oracle Text, previously know as interMedia Text and ConText, is an extensive full text indexing technology allowing you to efficiently query free text and produce document classification applications.

During one of my recent assignments, I stumbled upon some pretty nasty piece of SQL:


SELECT *
FROM source_table
WHERE
(
regexp_like(UPPER(column_one),  '.*(^|[^a-zA-Z0-9])'||:elem||'+($|[^a-zA-Z0-9]).*')
OR regexp_like(UPPER(column_two),  '.*(^|[^a-zA-Z0-9])'||:elem||'+($|[^a-zA-Z0-9]).*')
OR regexp_like(UPPER(column_three),'.*(^|[^a-zA-Z0-9])'||:elem||'+($|[^a-zA-Z0-9]).*')
OR regexp_like(UPPER(column_four), '.*(^|[^a-zA-Z0-9])'||:elem||'+($|[^a-zA-Z0-9]).*')
)

Ouch! Somebody wanted to find the rows from source_table that had the :elem bind variable as a whole word in any one of the four columns in our query.

This is a typical application for Oracle Text Indexing.

So we create the indexes:


CREATE INDEX source_table_cat_idx_one
ON source_table(column_one)
INDEXTYPE IS ctxsys.context;

CREATE INDEX source_table_cat_idx_two
ON source_table(column_two)
INDEXTYPE IS ctxsys.context;

CREATE INDEX source_table_cat_idx_three
ON source_table(column_three)
INDEXTYPE IS ctxsys.context;

CREATE INDEX source_table_cat_idx_four
ON source_table(column_four)
INDEXTYPE IS ctxsys.context;

and then rewrite the query:


SELECT *
FROM source_table
WHERE
CONTAINS(column_one,   '{:elem}', 1) >0
OR CONTAINS(column_two,   '{:elem}', 2) >0
OR CONTAINS(column_three, '{:elem}', 3) >0
OR CONTAINS(column_four,  '{:elem}', 4) >0

Still not good enough. This is where a cool little feature of Oracle Text Indexing comes handy: Datastores, or, more exactly, one particular type of datastore: MULTI_COLUMN_DATASTORE.

Here's how it works: in order to create one index for all the four columns we need to search in, we have to create a MULTI_COLUMN_DATASTORE preference:


ctx_ddl.create_preference('my_multi', 'MULTI_COLUMN_DATASTORE');
ctx_ddl.set_attribute('my_multi', 'columns', 'column_one, column_two, column_three, column_four');

This enables us to create the index like this:


CREATE INDEX source_table_cat_idx_one
ON source_table(column_one)
INDEXTYPE IS ctxsys.context
PARAMETERS ( 'datastore my_multi' )

So our query will be as simple as:


SELECT *
FROM source_table
WHERE
CONTAINS(column_one,   '{||:elem||'}', 1) >0

Upon examining the query EXPLAIN PLAN, you'll immediately notice that the whole search is done using the index we've just created. Another advantage of this type of queries is the fact that it enables us to use CONTAINS query operators, and thus make implementing a site-wide search a piece of cake.

Zend launches Zend Platform 3.6

2008-01-22T23:56:00.000+02:00

From Zend's press release:

Zend Technologies, the PHP Company, today announced the general availability of version 3.6 of Zend Platform™, the industry leading PHP web application server. Zend Platform 3.6 significantly expands the breadth and depth of its manageability, performance and clustering support for PHP, enabling organizations to better guarantee that their business critical web applications will meet end user’s quality, availability and response time expectations.

New features in a nutshell:

new caching options (in-memory caching, url-based caching, cache namespaces, client-side caching)
improved download server functionality
reproduce events on development server
event reporting options (e-mail, SNMP, XML, etc.)

I'm looking forward to playing around with it, especially the url-based caching (Zend Framework, anyone? :P), as well as the e-mail and SNMP reporting. God knows I know of a few websites that would use the performance and reliability boost Zend Platform provides.

custom Zend_Log_Formatter to display memory_get_usage( )

2008-01-21T15:58:00.003+02:00

I wrote a simple Zend_Log_Formatter implementation that adds the current memory usage to the output.

Source code after the break.


/**
 * Zend_Log_Formatter_Simple
 */
require_once 'Zend/Log/Formatter/Simple.php';

/**
 * Zend_Log_Formatter class to show mem_usage in log messages
 *
 */
class My_Log_Formatter_Simple_Memusage extends Zend_Log_Formatter_Simple
{
 /**
     * Class constructor
     *
     * @param  null|string  $format  Format specifier for log messages
     * @throws Zend_Log_Exception
     */
    public function __construct($format = null)
    {
     if ($format === null) {
            $format = '%timestamp% mem_usage=(%mem_get_usage%) %priorityName% (%priority%): %message%' . PHP_EOL;
        }
        
        parent::__construct( $format );
    }
    
    /**
     * Formats data into a single line to be written by the writer.
     *
     * @param  array    $event    event data
     * @return string             formatted line to write to the log
     */
    public function format($event)
    {
     $event['mem_get_usage'] = memory_get_usage( );
     
     return parent::format( $event );
    }
}

XML-RPC system.multicall method

2008-01-09T12:51:00.001+02:00

One of XML-RPC's default methods is system.multicall(). This allows the client to boxcar multiple RPC calls in one request. See http://www.xmlrpc.com/discuss/msgReader$1208 for details.

Implementation using Zend_XmlRpc is very straight-forward. For once, Zend_XmlRpc_Server already implements this method. On the client side, calling the system.multicall() method is done like this:

$client = new Zend_XmlRpc_Client('http://path.to/xmlrpc/server');
$client->getHttpClient()->setConfig(
                      array(
                              'timeout'      => 30
                           )
                        );

$server = $client->getProxy();

$response = $client->call(
    'system.multicall',
    array(
          array(
             array(
                   'methodName'   => 'setConfig',
                   'params'      => array('config')
                ),
              array(
                   'methodName'   => 'getConfig',
                   'params'      => array()
                ),
             array(
                   'methodName'   => 'getOtherData',
                   'params'      => array($sites)
                )
          )
       )
 );

Which Programming Lanuguage Are You?

2007-12-05T12:46:00.000+02:00

Which Programming Language are You?

Version Control with Subversion

2007-12-03T17:27:00.000+02:00

I've received many questions about SVN in the past, and I've had many questions of my own. Most of those questions have found their answer in O'Reilly Media's book Version Control with Subversion. This book is THE source of information about CollabNet's excellent open source version control system.

Leopard's Time Machine

2007-12-03T09:40:00.000+02:00

I've had a looong weekend to recharge my batteries, so I thought I'd give Apple Leopard's Time Machine a try. It looked really simple, and quite promising, to be honest. The interface was a blast to use and the snapshots were really fast and transparent. However, the thing works best for managing simple document backups rather than whole system level restore points. Having your system display "Library is a system folder and cannot be deleted or moved" when trying to restore to a past snapshot is pretty disappointing.

Zend_Cache and memcached

2007-11-28T17:18:00.000+02:00

Well, it seems that Zend_Cache supports memcached. Not a bad thing, considering who's using memcached these days. Look forward to using it in the future.

$this->isGeek();

2007-11-27T18:17:00.000+02:00

80% Geek

smartypdt

2007-11-27T16:02:00.000+02:00

http://code.google.com/p/smartypdt/ Eclipse plugin for Smarty. 'nuff said!

simple SimpleXML

2007-11-27T15:26:00.000+02:00

Well, i've done some simple XML stuff in PHP with (DUH) SimpleXML, and, being somewhat dissapointed with it's default asXML() output, wrote this little tid-bit of code to pretty up the output.

/**
* write XML contents to file
*
* @param SimpleXMLElement $xml         SimpleXML object
* @param string           $filename    name of file to write output to
*/
public function saveXMLFormatted( $xml, $filename )
{
  // write data to file
  // use DOMDocument->saveXML instead of SimpleXML->asXML because we want pretty formatting
  $doc = new DOMDocument('1.0');

  $doc->preserveWhiteSpace = false;
  $doc->loadXML( $xml->asXML());
  $doc->formatOutput = true;

  file_put_contents( $filename, $doc->saveXML());
}

Also, some people have complained that SimpleXML does not allow the deletion of nodes from the XML structure. That's not true, deleting a node is as simple as

unset($xml->node->child);

HTTP/1.1: Status Code Definitions

2007-11-27T15:24:00.000+02:00

http://www.w3.org/Protocols/rfc2616/rfc2616-sec10.html#sec10.3.3