First off, I think you might not be getting responses to this because you labeled it a 30-year software project, which isn't what you're asking really, you're asking how do you architect a product that will be maintained for 30 years.
Those are two VERY different questions. I came here expecting to answer not to expect to develop for 30 years.
Ok, now for my answer. There are a few things you'll want to look at.
First, I'd argue you build very modularly. You don't know what the product is going to need to do in 5 years, let alone 30 years. Building a product from independent modules will allow future developers to add, remove or update modules to reflect the times.
As far as languages, go with either what is popular today, or what you have EVIDENCE will be popular in the next 10 years. I say evidence because if you just hop on the next big thing (Go, Julia, etc) because at this point you don't know how difficult it will be to hire developers that are familiar with those languages in 10 years.
The libraries and dependencies answer is the same as the previous. Don't pick something obscure that you think 'might' be popular in the long-term, go with what's popular now or has evidence of popularity later. jQuery, Bootstrap, Rails, Node.js all had massive amounts of interest right from the start. However, if you can get away with not relying on a library, you should probably do it. If you need something now to ship quickly, but don't want it populating your project long-term, use it only in the modules that need it, and that way it can be easily replaced later (I'm doing this with jQuery now as Angular DOM traversal isn't great.. yet).