It seems like we've surfaced a bunch of short- and long-term issues here that are all getting mashed together. The short-term issues are somewhat easier: For example, if Jamie thinks he'll be able to get AOLServer 4 final running on Windows, that's clearly the best short-term solution for Windows compatibility. (In fact, it may be the *only* short-term solution for Windows compatibility.) Everybody also seems to agree that using Java packages in the short-term for stuff where Tcl code doesn't exist isn't a bad idea. This is a bit trickier, since at least some here are concerned with dependencies, but the service-contract approach sounds like a strategy that people agree will work for handling this.
I'd like to float one more solution that I'm explicitly tagging as short-term. Regardless of the long-term viability of *replacing* AOLServer with nstcl on FastCGI, might this be a shortcut for getting mod_AOLServer on Apache? If nstcl has all the necessary features, and if it can run without too much work using mod_tcl, it seems to me that we could get Apache compatibility relatively quickly. We could at least try to benchmark the performance...
Longer term questions are murkier. Can AOLServer be made more portable to a range of operating systems? Can it be made more marketable? Can it be replaced? If so, with what? Is Java a part of the long-term solution or not? And if so, how? Rather than argue these issues with no data, we should lay down a plan for exploring our options. Advocates for each of these paths should come up with a plan to explore the options and see if objections can be answered. The most attractive plans will most likely be the ones that attract resources to carry them out.
