Maybe it’s because I’m speaking at the Zend Conference in October, maybe it’s because it’s a frequent topic of conversation in this space, but lots and lots of folks have written in, called, or IM’d me about the news from last week concerning Evans Data Corporation findings vis a vis PHP to get my reaction. In case you missed it, Martin’s got the details here, with this quote pretty much summarizing the findings:
“PHP, Perl and Python use on a global basis peaked one to two years ago and has started to decline based on a number of factors. This decline is more exaggerated in EMEA and APAC than in North America,” said John Andrews, Evans Data’s Chief Operating Officer, in a release. “One of the key factors to this loss of developer mindshare has been the inability of these languages to penetrate the enterprise space.”
So the question posed to me has been what do I think? Well, before I answer let’s back up and look at the firm making the claim.
I don’t know the folks from Evans, but I can tell you that in our corner of the industry they’ve carved themselves out a decent reputation as trackers of developer adoption. To the best of my knowledge, they are aboveboard folks, not dipping into the “commissioned” report waters. All this should be taken to mean that I know of no good reason to discount the findings as the result of bias, or a paid for opinion. It should, therefore, be given fair consideration.
So in going ahead and considering it, I’m have to say that I’m pretty bewildered. I’ve had my disagreements with Evans numbers before – notably around the declining C# numbers that Java advocates cite when I bring up the topic of Mono – but they weren’t even in the same ballpark as the PHP (and Perl/Python) report. The disparities that I’ve observed in the past with what I see in my coverage versus what Evans produces have typically been slight, but this one more or less contradicts everything that I’ve observed in the last 12 months or so.
As Martin notes, the “inability of these languages to penetrate the enterprise space” bit is particularly strange, given the recent deals with first IBM and then Oracle. While the level of commitment for those two is on the partnership level, it’s clear that the driver for the announcement was, in fact, enterprise traction. When the announcement of the IBM/Zend partnership was originally made, it was often portrayed as “now PHP is enterprise ready,” which in my mind misreads the original situation significantly. The fact is that IBM and Oracle were both seeing a lot of internal enterprise projects running PHP, and found that in many cases these projects were skipping DB2 and Oracle in favor of alternative open source databases that were more integrated with the scripting language. Once this market reached a certain size – meaning large – the decision for both parties was relatively simple. Support a new and growing market, or cede it to competitors. Sun’s getting in on the PHP bandwagon as well via the efforts of Bryan Cantrill and Wez Furlong.
PHP’s not alone, either. Evans makes similar, if less dramatic, claims about the decline of Python – which makes the Microsoft hiring of Jim Hugunin another questionable move. BEA’s also on the wrong track with its plans. If they’re taken at face value, these numbers mean that the a fair sized selection of the largest software firms on the planet have misread the market. Either that, or the survey’s a fluke.
To look critically at the future, however, is the outlook for PHP nothing but sunshine and flowers? Nope, as Ruby is positively roaring out of the gate. It’s still a speck marketshare-wise, but the language combined with the Rails framework is proving remarkably compelling for developers seeking lighter weight approaches to C# and Java. But Ruby certainly is not an explanation for Evans numbers, because if PHP hasn’t penetrated the enterprise market, Ruby’s never even seen it.
Let’s accept for the sake of argument that Evans is correct, as an exercise. I then have one simple question: what are developers turning to as an alternative? Surely Evans isn’t arguing that the need for lightweight, highly productive application environments has departed along with the willingness to consider PHP/Python/Perl? If the Evans answer is Java or C# I’ll be very skeptical of the claims, and as I’ve already mentioned I doubt PHP’s suffering at the hands of Ruby within the enterprise. It may be that Evans answers this in the report, but it looks like that’s available for paid subscribers only.
In a nutshell, I cannot offer an explanation for the trend that Evans is reporting b/c it does not reflect any of the evidence that I’ve observed, be that anecdotal or corporate roadmaps. Whether this is a “Dewey Defeats Truman” moment or not remains to be seen, but for now I’m going to take the report with a grain or two of salt.
 There has been, however, a lot of interesting speculation about what it would take to get there. For the fans of Ruby that feel I’m dismissive (despite the fact that I’m learning the language now), have patience – I’m watching closely, and will be happy to revise my story if I see things changing.