> Seeing all the backward compatibility guarantees, my personal feeling is that > if a "breaking change" doesn't break the existing libraries in nimble, then > it's probably fine. Sometimes a fix is important, and I hope it won't deter > the innovation and improvement of the language. Comparing the percentage of > libraries that get broken after a change, could be a good measure if it makes > sense to incorporate it anyway.
We have a selected set of "important packages" that are now part of our test suite, so to some extend we already do that. However, most code that is written is closed source code. So it's a very good indicator but still we should be cautious. I'd like to be Nim as stable as the other players and if that means to live with some design warts for longer than desirable, I'm willing to pay the price.
