I think it's doomed to fail. If I value stability over features, I will obviously take C89, C99 or some C++ standard.
Besides, we truly need some revolution in system languages, I believe. ATS, F/Low and Rust are really good first tries, but we are not there yet in terms of soundness/completeness optimum. So it will take time and many attempts to develop some truly safe and simple language with resource tracking, generalized compile time computations (i.e. effect/totality tracking and dependent types) and so on.
If I were looking for a 100 year language, I think I might go a little higher level and target Lua (say 5.1 so you could plug in LuaJIT and/or NeoVim plugins). As a dead release, it is already at a situation of no feature churn. The VM is minimal and there are implementations in many other languages.
> Hare instead aims to distill the as-of-this-decade state of the art in proven systems programming language design idioms to produce what we believe is the most robust language design right now.
This quote could really use a secondary blog post outlining why they believe this is true relative to other languages. Putting aside where it stands on a relative basis, there are several decisions that they make that I personally think are directly in opposition to the absolute goal of robust program design. One specific example is that Hare supports global variables which is pretty hard for me to accept in a greenfield language.
If they want to last 100 years, I would suggest that they go far smaller than they have. It is clear that they could self-host the Hare compiler with a much smaller subset of features than are in the language spec. That smaller Hare would be much closer to the true essence of Hare and would make it much more likely that they would create something capable of enduring for 100 years. It's a great goal but I think they have a lot of work to do if they have hope of achieving it.
Languages need to evolve because the needs change over the time. It's either evolving or the users will have to switch to another language eventually.
The goal of my language[1] is also to be unchanging after the 1.0 version. However the language is extensible in many ways therefore not hindering the ability to evolve while still getting the benefits of the super stable base that you can build everything on.
The first 3 things weren't a consideration in any of the successful languages for which I'm familiar with the genesis of. The closest is probably Java, but that language changed a lot post-1.0
Besides, we truly need some revolution in system languages, I believe. ATS, F/Low and Rust are really good first tries, but we are not there yet in terms of soundness/completeness optimum. So it will take time and many attempts to develop some truly safe and simple language with resource tracking, generalized compile time computations (i.e. effect/totality tracking and dependent types) and so on.