Just the facts ma'am

PHP 5 was a significant leap in technology. It bought us a wealth of new features, and anyone who has used it in production will appreciate the direction it took. Publishers and magazines went into overdrive, with book after book appearing on the virtual shelves. 'PHP 5 with Objects', 'PHP 5 with Design Patterns', 'PHP 5 and Security', the list of topics goes on. The Zend site published articles on ways to harness the new features and move over from PHP 4. The main PHP magazines focus mostly on version 5 as well, especially as that is where the most exciting development is happening. So from the outside things look rosey. But look at the actual facts and the picture is not so pretty.

Just 4.2%

The Nexen.net PHP statistics bot polled over 9.5 million web servers in September 2005. Their results? That PHP existed on over 46% of the servers polled. Great news indeed until you dig further and see that of those servers with PHP on, only 4.2% were running PHP 5. An amazing 94.5% were running some flavour of PHP 4 (the remainder being PHP 3).

Mailing List Chatter

A topic that a lot of people still ask on the PHP General Mailing List is 'Where can I find a host that supports PHP 5?'. While it is true that there are now many more supporting hosts than we had a few years ago, the fact that even today people are still needing to ask this question highlights the fact that it isn't yet common-place for web hosts to deploy servers with PHP 5 as standard. A large number of those that do are what I would term 'Developer Friendly' hosts, who have packages catered more towards developers than Joe Public. Others try and shoe-horn support in via a CGI version of PHP alongside the PHP 4 Apache Module.

Release Rates

Looking at the release rate of updates for PHP 4 and 5 between 2004 and 2005 you will see that PHP 4 had 9 updates (from 4.3.5 to 4.4.1). PHP 5 had 8 updates (5.0 to 5.1.1). Each upgrade was of course different, some being pure security fixes, others bringing in exciting new enhancements such as PDO into the PHP 5 core. I mention this fact for the simple reason that it is this duality that isn't helping in contributing towards PHP 5 adoption rates. While it is of course sensible to keep PHP 4 current, given the massive installed userbase for it, doing so sends out mixed messages to the community at large. Both to developers and web hosts.

In a stark contrast if you look at the change-over from PHP 3 to PHP 4 it happened in a mere 5 months. PHP 4 debuted in May 2000 and PHP 3 gave out its last release breath on October 20th of the same year. Think about that for a second - in just 5 months version 3 had all but been stomped on by the mighty v4 juggernaut. We had all of the same 'backward compatibility' issues with 3 to 4, hosts had the same issues with them not working together easily, and yet everyone just took it for granted and made the leap. The baton-passing from 4 to 5 still hasn't happened yet.

You can quite rightly argue that perhaps this speed of change was due to the PHP usage rates back in 2000, with only 2.5 million domains running PHP. Whereas today the number is closer to 22 million, has this popularity lead to its own sluggishness in uptake?

Clients drive demand

Even for experienced PHP developers, who love working with the features that PHP 5 provides, it isn't always so easy to make that change. In 2005 I personally managed to start work on only 1 project that used PHP 5. Everything else was maintaining legacy code running on PHP 4, or creating code for hosting environments / servers that would only run PHP 4. So as much as I personally would have liked to use PHP 5, I was unable to. The restriction came from those who control the purse strings. I am sure I cannot be alone in this situation, and I would imagine it reflective of a large chunk of the development community.

Is talk of PHP 6 helping or hindering?

One of the most viewed articles on my blog is that covering the round-up of the PHP 6 development talks. Although we are far far away from version 6 (indeed I would hazard a guess that we won't see a stable release until 2007), it is already a hot topic. One of the things I noticed was that people liked the finallity of the proposed version plans so far. The way in which 6 is doing-away with legacy issues such as Register Globals. On the surface alone it appears to be a far more 'solid' version. Not in the 'stability' sense, but in the way it will hold itself and make some bold and courageous moves. I dare-say moves which some of us had hoped PHP 5 would have done. Perhaps there is an element of us that thinks 'I'll just wait for that version', in a similar manner to the way in which someone might wait for Windows Vista to ship before upgrading their PC/OS.

Reasons why...

Let's take a quick look at some of the reasons why so few people have moved from PHP 4 to 5:

1. Unless you are a hardcore PHP developer, there is no real compelling reason to need to change
2. Web Hosts have perfectly working PHP 4 environments, they won't change unless their clients drive it
3. Most 'out of the box' Linux/Unix distros (especially those used for co-location) still have PHP 4 installed by default
4. There are thousands of PHP 4 specific tutorials and resources out there
5. Some of the most common web applications still only run on PHP 4 (phpbb springs to mind - we'll leave the relative merits of this bug-fest software for another day, but you cannot deny its popularity)
6. Publishers are still churning out PHP books related to version 4
7. Clients often want you to support existing legacy systems rather than re-build from scratch
8. It just works

What is required to combat most of the above is a 'killer app'. Could the Zend Framework be the saviour that PHP 5 requires?

A Knight in Framework armour

The differences between 4 and 5 are significant, but only to 'those that know'. Out of the 22 million domains running PHP, how many of them are likely to be professional developers? I would hazard a guess at a very small percentage indeed. Therefore the technical differences are of far less importance and value to Joe PHP Public. Right now they are probably well aware that PHP 5 is here, and here to stay. But they keep seeing version 4 updates, they keep seeing updates to the applications they may have downloaded and installed onto their web sites, and right now there is no real benefit for them to move. But could a killer-app change this mindset? I will wager that this will be the case, but I am ever the optimist!

I see the Zend Framework as being intrinsincly important to PHP 5's adoption rate. Yes there are many other frameworks out there already, and I don't wish to belittle the work and energy that has been put into creating them, but at the end of the day anything that gets an official Zend stamp on it is going to become popular. They simply have both the resources and the developers mindshare to make it happen. Plus they've got some seriously talented PHP developers onboard to ensure that the framework itself is both practical and well featured.

There are masses of other PHP frameworks out there, lots of which utilise the features of PHP 5 (or are PHP 5 specific). But it would be fair to say that none of them have really captured any significant kind of 'market'. If you start a thread on the PHP General Mailing List asking 'which framework should I use?' and get 20 replies, you can bet your bottom dollar that you'll be given 20 different frameworks to look at. You need to invest a lot of time, research and patience evaluating which is right for you. And then if, after you've finally picked one, development on it suddenly stops - you'd feel slightly agrieved. This is why I believe that a large percentage of PHP developers choose to 'roll their own' until a better alternative comes along, and just perhaps the Zend version will be the one.

Some things can be taken as a given: We will see tutorials on using the Zend Framework spring up, both in PHP magazines and on web sites. Given time, we will see books specifically about the Zend Framework. I would imagine that it is possible we could even see the framework bundled with PHP itself, at least I hope this will happen because it will help to change peoples perception of PHP 5. You won't just get a language 'out of the box', you'll get a complete development environment. That is a significant distinction. Right now Zend might be at pains to enforce the fact that they don't want to 'compete' with other frameworks, and that they can all live harmoniously. But to be honest I'd rather they took a more direct approach, a less ambigous one - and did indeed aim to be the dominant, first-choice framework for PHP.

Perhaps other framework authors are realising this already? Only the other day Davey Shafik announced on his blog that he was dropping work on Cerebral Cortex, his own framework. This coming only a few months after a Pro PHP Podcast stating the benefits of it. One of the reasons he cited for this decision was the upcoming Zend Framework, and I for one don't blame him. He may be the first high-profile PHP developer to have done this, but he surely won't be the last.

A quick sideways glance into the Ruby world highlights the importance of the framework (Rails) in the adoption rates over there. RoR is a 'killer app'. If you take one look at Ruby itself as a language, there is virtually nothing about it that appeals. It has a unique, bizarre and frankly hard-to-read syntax. Drop Rails into the equation however and suddenly you've got something worth investing a little R&D time into. The combination is what makes it popular, not the induvidual components.

The Zend Framework may be more important to the life expectancy of PHP than perhaps even they realise. Right now PHP 4 is the dominant force, riding high. There is a worthy and powerful successor in the form of PHP 5, sat twidling its thumbs, just waiting to leap into popular use. All it needs is some kind of catalyst. Fingers crossed that we're on the cusp of it, because PHP 4 cannot sustain its massive popularity indefinitely. The Internet is just too fickle a place for that.

Data Sources

PHP Version Chart: http://www.corephp.co.uk/docs/php_release_chart.php
Nexen.net PHP Stats: http://www.nexen.net/interview/index.php?id=51
Netcraft PHP Usage Chart: http://www.php.net/usage.php
Amazon.com Publication List 2005