I remember an article on HN a couple of years ago about the temptation to build a software system on specialty, "interesting" languages or libraries that seem adapted to your specific needs and desires.

But the downside is that it costs a lot to find people who can code in that language, there's little community support, and you run into equally specialized problems that are hard, expensive, and take a long time to fix.

It seems there are lessons that parallel this house experience.