<blockquote><i>But until these obstacles are actually overcome, I think the community would be wise to actively investigate long-term alternatives, even if those alternatives are clearly technologically inferior...today.
</i></blockquote>
The problem is that the suggestions are most likely to be technologically inferior <b>forever</b>. As Michael Cleverly points out , a Tcl-based solution is never going to approach the speed of a C-based solution (AOLserver). nstcl today solves perhaps a quarter or a third of what would be needed to support a poorly performing FastCGI solution.
<p>mod_nsd will never be part of the Apache mainstream, leaving us with maintenance issues similar to our having to support AOLserver on windows. Note that we <i>have</i> AOLserver on windows in preliminary form while mod_nsd hasn't been maintained for years, now, and moving it to Apache 2.0 will be a big job. An argument that supporting mod_nsd will be a bigger headache than supporting AOLserver on Windows is easy to make, given that it fits today's situation.
<p>It's tempting to try to fix marketing problems by changing technology, just as it's tempting to fix technical problems by changing one's marketing approach.
<p>ArsDigita tried to fix their marketing problems by dumping ACS Tcl and rewriting it in Java. ArsDigita no longer exists. The problem wasn't the technology, it was the fact that the team in charge at the end probably couldn't sell snowcones in the Sahara.
