Thu, 29 Jul 2004

Reliable software and other lies

At recent Humbug meetings, Andrae has teased me with references to Erlang—but so far I’ve been strong and avoided learning more. But now we have Martin drawing our attention to a Usenix paper about crash-only software and Andrae has taken the opportunity to direct us to Joe Armstrong’s PhD thesis Making reliable systems in the presence of software errors as a counter-point to the Usenix paper.

So I had a look at Armstrong’s paper—it turns out to be a 295-page PDF file that describes a large Ericsson project that resulted in the development of the Erlang language as its implementation tool. I haven’t read the whole thing, but Armstrong’s web site took me to a bunch of interesting things including the Erlang documentation and to the home of Yaws—yet another web server—written in Erlang and with a facility to embed Erlang code in web pages, much the way PHP works.

I thought I might try Yaws and Erlang (although I really want to focus on Lisp and Scheme this month), just to give me a way to avoid having to play with Apache. But then I was worried that I might have to abandon a couple of important legacy applications I use that are written in PHP. Fortunately, I was clever enough to read the Yaws manual and so discovered that it has support for PHP as well. So, I plan to setup Erlang and play with Yaws.

I had been toying with the idea of writing a web server in Lisp or Scheme (and might still do that), but this one might save me some work in the immediate future. And anything that lets me do dynamic web pages while avoiding Apache just has to be a boon.

Why am I so down on Apache? Aren’t they the good guys? Well, no they’re not the good guys now, even if they were once. It seems that the Apache project has really taken on the Microsoft approach to software—or perhaps I should now call that the Microsoft/Java approach. In essence, these systems seem to be propelled by a desperate need to build bigger and more Byzantine software artifacts, regardless of their utility and despite the clear evidence that this way madness lies. I could go on at length on that topic, but I’m tired and so I’ll stop there for now.

Posted at: 18:00

Path: /software/discuss   |  permalink