F#: Two Remarkable Years

>>F# can also claim to be the first .NET language to support generics.

This is pure bollocks. Eiffel <www.eiffel.com> has generics and was ported to .Net long before F# even saw daylight.

Nice article - thanks for the mention.

Eirik - the team at Microsoft Research Cambridge invented .NET generics and used an early version of the F# compiler extensively in early testing. We designed and implemented the generics support for the F# and C# compilers and did essentially the complete implementation in the .NET Common Language Runtime, which then became part of the product. Of course the design of generics in Eiffel formed a significant part of the design backdrop.

I guess you could say our modified ILASM assembler was the first language to target generics, followed by the ILX assembler and an early (now discarded) attempt at a "Haskell .NET". Then came early versions of the C# compiler and an early version of the F# compiler as we completed various bits of each (the very first 'hacked up' F# compiler was in 2002, though the language itself didn't really take shape until much later). Early versions of all of these were used extensively in the testing of .NET generics.

Cheers
Don

F# is a very promising and fascinating project. Looking at the assemblies generated from F# was quite a nice experience. However, last time I checked there were two major problems that kept me from looking further into it:
- F# does not support meta-programming, one of the more interesting features of OCAML (via caml4p). .NET is really lacking here, and it would be great to see a language like F# provide _something_ in this area.
- The license is practically unusable for commercial purposes. The need to link statically prevents you from creating F# library under any realistic circumstances.

Stefan

Don -- I'm not disputing the info in the underlying article (found here <research.microsoft.com/fsharp/about.aspx>).

I'm fully aware of the research that's been (and is) going on and it's implementation in e.g. Rotor.

I was merely pointing out that the statement "F# can also claim to be the first .NET language (my emphasis) to support generics" isn't true. For it to be true, it should have said something like "F# can also claim to be the first .NET language to utilise CLR-generics" or something similar. I would still claim that the first .Net language supporting generics was Eiffel.

That you added support for generics was great news. If only the CLR would support MI as well :-) (perhaps you could pick up a few ideas from Eiffel's implementation of MI).

As an end note, it is nice to see functional programming being brought into the mainstream and it is interesting to see that it is Microsoft who's doing so.

Regards,

Eirik
"If I can't Eiffel in heaven, I won't go"

"F# does not support meta-programming" -- see tomasp.net/articles/fsclassmeta.aspx and research.microsoft.com/~dsyme/papers/ml03-syme.ps

Alex, while this looks interresting, i don't think it goes far beyond C# 3.0's Expression<T> construct. I want to be able to code without quotations, and have some metaprogramming rules applied to the code afterwards. I don't think quotations would solve any of my real problems.