Since they are usually written by non-professional programmers, this is not good
Well, that's the problem, isn't it? My point when dismissing lua isn't that lua is a bad scripting language. It works, and it's been developed for years and have a semi-large following. My point is that since the scripting language we use in the mud is crucial to the uptime of the mud, we're not likely to want to have to start over on the hardening.
You mention that a bad trigger can crush the server - but we've iterated over that situation for years, and there are now fewer scripts that will lock up the server than there was when we started. This process would have to start all over again with a new scripting language.
One might argue that since lua is a separate language, it is easier to embed it and limit the scope of the sandbox. At the same time, this introduces unknowns - we have the source code of dg scripts readily available and can fix it if we get a bug report. With lua we'd either have to add a library dependency (which we'd very much like to avoid since this is notoriously difficult cross-platform) or we'd have to add some external source code as part of the build, and keep it up-to-date over time.
I agree that dg scripts has a limited audience. That does not necessarily mean we should ditch it for the sake of it.
