Ada has been held back primary by an image problem that traces back to the high cost and poor performance of a lot of early Ada 83 compilers. Ada adoption has never really recovered from that despite its many technical advantages, and despite the low cost and good performance of several current compilers.

The GNAT Ada compiler, always open source and quite good, has been freely available since the 1990's. It has been part of GCC since about 2003.

There are plenty of open source Ada projects on GitHub and other places although not nearly as many as some other languages.

The Ada ecosystem is mature and complete, particularly the GNAT related tools supported by directly or indirectly AdaCore (https://github.com/AdaCore and https://alire.ada.dev/).

The language evolution has been stable and is still on-going. I have worked primarily with Ada for 30 years. I still work on new Ada projects on a mid-sized team. Most of us just don't participate in forums like this.

> early Ada 83 compilers

IIRC, in response, DARPA (et al) did invest in compiler research.

> adoption has never really recovered

Ya. Timing. There's a brief window of opportunity for new languages (ideas) to catch on before the horde of "worse is better" solutions overwhelm the field.

For people who were actually interested in using Ada, the early poor tooling was the main impediment. However, I think the bigger issue is that it was moving directly against the cultural headwinds at the time.

The immediate response I heard anytime Ada was mentioned was that it was a designed-by-committee language[1] that couldn't even be fully implemented due to a theoretically impossible specification[2]. It was made by a bunch of bureaucratic stiffs and was all about constraining the developer with stupid rules and bogging them down with verbosity. It was contrary to the freewheeling nature of the PC developer culture that sprung up in the 70's and continued through the 80's, and then evolved into the dot-com developers of the 90's and 00's.

It took decades of wandering through the deserts of "Real Developers don't write buffer overflows" on one end, and "Performance doesn't matter, and a sufficiently smart compiler will provide it anyway" on the other to get to the point where mainstream developers wanted a language that combined the safety of high-level languages with the control of low-level languages.

[1] This is false, it was selected in a contest with each entry developed independently.

[2] True but overrated