Abel Braaksma

Re-reading this thread, I can see essentially the following, relatively cheap (I hope) areas for optimization by the compiler: 1. Each case has same type, or no type (empty case):...

This is related: https://github.com/dotnet/fsharp/issues/7826 (but not the same, that issue is caused by incorrect preview feature parsing).

Yes. from F#'s standpoint, these are just `longident`, and while they have special meaning in some scenarios, they are not illegal. Likewise, if you do have an active pattern, you...

@charlesroddie, like the `IndexOf` methods for strings, I think it's reasonable to return `-1` for empty lists, arrays, sequences. Btw, there was a suggestion here to implement this for `Map`,...

What happens here, if you look at the compiled code, is that it gets compiled into nested if-statements for situations that are too difficult for the compiler to rewrite more...

@cartermp, thanks for mentioning the 3-SAT problem, reading up on that gives some hints on how this problem can be problematic. Though I don't think it is unsolvable, because DU...

> Especially if @abelbraaksma is correct in that I like to think I am ;). The code above can easily be tested, just check the compiled binaries (and watch the...

> The majority of pattern matching code compiles well. Issue #5212 suggests that exponential growth can happen really easy and diesn't require record fields. Looking at the IL I don't...

You'll need to use Razor Generator, see the readme: https://github.com/HangfireIO/Hangfire?tab=readme-ov-file#building-the-sources. I have not done this myself, though, I haven't touched Razor in years...

@konst-sh I think this language suggestion is about type definitions in F# code itself, where you would allow nested types (like we allow nested modules). It'll need a volunteer for...