I used to write a lot of projects in PHP when I was working for creating a Web interface to something. However, I recently have spent some time learning other languages—C# and Java being two of them—and am beginning to realize that I think I would rather use one of them for creating anything larger than the smallest of projects anymore. Of course, their hosting environments differ widely, but I am finding that while I don’t know how the languages compare with the Apache PHP module in terms of performance, I would find the way that I think in terms of programming to be more easily expressed in either of those two languages, compared to PHP.

I used to think that language features found in the C family—notably, strongly typed variables—were nothing but a painful nuisance. At least, I thought that way until probably a few weeks ago, when I started work on a PHP project after spending a lot of time working in Java for a class and C# out of my own interests. And I realized something about languages that use strong variable types: Languages that require strong typing of variables might be inconvenient sometimes, but they build a lot more documentation into the program, simply by requiring strong typing. In a language such as PHP, you simply cannot tell what the variable’s type is by looking at the variable name. If you follow conventions (and not all people do) for naming your variables, you might be able to create little “hints” to their types, like $isAuthenticated for a boolean value or $startedWhen for a timestamp, date, time, or other similar type. The Hungarian notation might work for PHP, too, but I have always found that rather hard to read, particularly in certain situations. Instead, properly written code in a strongly typed language makes it very clear what the variable is, does, and may contain. Very fascinating at least to me, simply because I have always disliked what I perceived to be inflexibility in strong data types for variables.

Of course, this isn’t the only reason that I am finding I would want to work with one of those two languages. I absolutely love the idea of generics (yes, I know that they’re in C++… I never did play with that language enough to “get it,” apparently). I like the simple power of the languages (and their associated class libraries)—you can do things like create a dynamically sized array in a way that feels natural and non-hackish, using a type of generic called a “collection”, which goes a long way towards readability (and as strange as it sounds, that power partly comes from the extra documentation provided by the strong types used with them). Given that “code is read much more often than it is written”, it makes sense that features that reduce the amount of programmer time like this—and are readable—are things that I find useful.

There is only one real downside to both C# and Java, when compared to a language like C: They are far, far slower. What they essentially do is transfer programmer time to system processing time—depending on an application’s lifecycle combined with many applications running at the same time on a system, the system processing time could indeed be more than the time a programmer saved writing the program to begin with. C# tries to mitigate this by compiling the program’s bytecode “just-in-time” so that native code runs, and code that executes repeatedly gets a benefit from this. Java does the same thing with HotSpot, though only in certain areas of a program as the Java VM figures is necessary while analyzing the program during its execution. Still, they are slower and use more memory. PHP probably is, compared to C, as well. I think it would be interesting to compare PHP against C# and Java in a Web environment, though I have not done that just yet since there is no really easy way to do it equitably that I am aware of.

Anyway, how does this pertain to Web programming? I am beginning to think that I should use PHP for only small things that are not required to be very powerful or have long maintenance lifecycles. Why? There’s a ton of PHP programmers out there, to be sure, and PHP runs “out of the box” on a lot of services, such as NearlyFreeSpeech.NET, whereas something like mod_mono or Tomcat do not. You might be required to manage more of an application server with the latter choices, while the former ones are more flexible in most installations and you can manage just the application and leave the server management to a crew of people that want to do that. Where, then, is the benefit?

It would appear—at least to me—that the increased investment in initially setting up a Web application server (and maintaining it) with an environment for either C# or Java would be paid off in that wonderfully limited resource known as programmer time—as well as frustration. Personally, I hate re-inventing the wheel. And while there are resources for PHP that will let you do that, sometimes they require extensions to be installed or—often—the use of ad-hoc class libraries written for PHP, such as those found at the PHP Classes Web site (which, IMHO, is a poorly designed site to begin with). Unlike PHP, C# and Java both come with very large, powerful, and more importantly, standard class libraries, which is something that greatly helps programmers with both reading and writing code. There is a lot of support out there for using third-party class libraries (which you can find all over the place), and they are supported by a lot of third-party programs with interfaces, too (such as MySQL and PostgreSQL, and others).

All told, learning these languages has got me wondering if I shouldn’t use one of these languages instead. While I’ve really only scratched the surface of learning about many of the other things (such as the details about creating Web applications in these languages and hosting them), it seems that overall, there are less headaches to be had and code can be written faster, debugged faster and more efficiently, and there is more convention built into the language to govern how things are read and written by other people.