Posting Arc code on Arc Forum is a good way to be sure that someone else can run the code, even when we're talking about speculative language design. I get the impression LtU has the same effect going on, but with Haskell as the assumed common language. Before, we've wondered what kind of forum features would be good for Arc Forum. Things like having runnable, searchable examples so our code doesn't get lost to history. But here's a spin on that: Suppose the forum software and its lingua franca language are being designed at the same time. What forum features do we want? How should that affect the language design? If you're reading this and you're already designing a language, do you already feel it's exactly the right language for this task? I'm doing open-ended language design recently, without a single target language in mind. But I'd say I haven't considered any target language that would be up to this task! Forums are not flat, the way the space of open source software projects tends to be flat. Some posts are responses to others, so they may benefit from tools to import or tweak the responses that came before, tools to demonstrate failing unit tests for previously posted code, and tools to demonstrate that new code passes previously posted unit tests. There probably needs to be some amount of hackish metaprogramming support, for the purposes of meta-discussions about improvements to the forum language itself. There's pressure to turn integrate this with a wiki IDE, but I personally consider that to be outside the scope of this idea. If we go too far down the path of generalization, we're not talking about forums anymore. |