2016-04-20 19:52 GMT+02:00 Roman Cheplyaka <roma@ro-che.info>:

On 04/20/2016 08:06 PM, Giacomo Tesio wrote:
> Is this a language issue?
> I don't think so... but apparently, despite the abundance of language
> hackers in the Haskell community, nobody still tried to prove that an
> interpreter for the core Haskell language can be written in a reasonable
> amount of C code.

You are contradicting yourself. Mark P Jones and other "language
hackers" who contributed to Hugs have already proven that.

No contradiction, you are just misreading the statement: P. Jones proved that for Haskell98. For Haskell2010, it's still to be proved.

That no-one seems to be willing to maintain Hugs may indicate that there
aren't as many use cases as you claim.

Maybe.

But, as always in engineering, it's a matter of economics.

> Thus, to my money (and admittedly for my own use cases), if somebody
> renew the interest around a simpler Haskell implementation, he's going
> to have a really *positive *effect.

Would you put your money where your mouth is? I'm sure you would find
someone who could maintain and improve Hugs for you.

Sure, give me an estimate of the cost! I will seriously evaluate the investment.

The requirements of the update are:

- support Haskell2010 without any extension

- simple and portable C code, possibly C99 but at least GCC

- statically linked (possibly newlib, but I can replace the libc later)

Actually to evaluate the RoI I need some statistics about the Hackage packages that such renewed Hugs would be able to run:

- how many packages in Hackage are currently maintained? how many of them did exist 2 years ago? how many packages were maintained 2 years ago?

- how many maintained packages conform to Haskell2010 without using any extension?

- how many maintained packages use FFI (requiring libraries that should be linked with Hugs too)?

- could such interpreter run cabal?

Giacomo