[Haskell-cafe] Re: Is www.haskell.org down?

[Jon Harrop]

On Thursday 14 August 2008 19:52:06 Ketil Malde wrote:
> Don Stewart <dons at galois.com> writes:
> >     http://article.gmane.org/gmane.comp.lang.caml.general/43430
>
> Well, to be perfectly honest:
> | > What are the advantages/disadvantages when comparing OCaml to Haskell?
> |
> | Techically, predictable performance, high performance on x64, tools like
> | camlp4 and language features including records, functors, polymorphic
> | variants and structurally-typed objects are probably the main advantages
> | of OCaml over Haskell.
>
> I think we can match language features with classes, concurrency,
> GADTs, STM, and so on.  Predictable performance is a point, especially
> memory-wise. At least, I sometimes get surprises when a usage pattern
> causes too much or too little laziness.  GHC uses a *lot* of memory in
> general, and in particular on 64 bits.

Yes. Compilation speed as well.

> I belive OCaml has a simpler semantics, and thus it is easier to
> optimize performance, especially for a beginner.  Decent programmers
> get decent performance with Haskell, see e.g. the language shootout.

Isn't Darcs an obvious counter example?

> | Non-technically, OCaml has more well-tested libraries, more practical
> | books (e.g. my own), more high-profile open source projects (e.g. FFTW
> | has millions of users)
>
> I'm not so sure about the library thing, it seems that Haskell has a
> bigger community,

What gave you that impression?

According to the Debian and Ubuntu package popularity contest results, OCaml 
currently has 10,635 registered installs compared to 6,606 for GHC. Moreover, 
this ratio has been constant for several years now.

> and while some libraries are less maintained than 
> one would wish, some are pretty solid.  Clearly, OCaml has been used
> in some serious settings, although I don't see why the number of users
> of one particular application should be significant to me.

There are data for more applications, of course:

Debian and Ubuntu registered installs
-------------------------------------
184,574: FFTW       ( 14,298 lines of OCaml)
 12,866: Unison     ( 23,993 lines of OCaml)
  7,286: MLDonkey   (171,332 lines of OCaml)
  4,365: Darcs      (  3,939 lines of Haskell)
  4,066: FreeTennis (  7,419 lines of OCaml)
  4,057: Planets    (  3,296 lines of OCaml)
  3,465: HPodder    (  2,225 lines of Haskell)
  2,965: LEdit      (  2,048 lines of OCaml)
  2,822: Hevea      ( 11,596 lines of OCaml)
  2,657: Polygen    (  1,331 lines of OCaml)

That is 221,293 installs of popular OCaml software compared to only 7,830 
installs of Haskell software.

And 235,312 lines of OCaml code tested by many users compared to only 6,164 
lines of well-tested Haskell code.

> | There are far more practically-minded people in the OCaml community
> | and that is reflected in the much larger number of commercial
> | products that use OCaml.
>
> This almost reads as damning with faint praise.  Do we have the
> educated, smart people, and a larger number of open source products?

The above statistics certainly that OCaml has four times as many popular open 
source projects as Haskell.

> | Tangentially, OCaml has a successful relative in F# whereas
> | Haskell's close relatives are extinct.
>
> Or you could say that OCaml is one member of a family of languages,
> each slightly different, while Haskell managed to unify the
> ecosystem and community around a single standard.

Or you could say that the only popular Haskell is GHC's proprietary dialect.

Either way, I think it is clear that OCaml has more users than GHC and 
OCaml+F# have far more users than all Haskell dialects combined.

> It would bear pointing out that just about every new, cool feature in
> emerging languages seem to come from Haskell.

Like F#'s active patterns?

> | nobody has ever done anything significant using Haskell
>
> He is entitled to his opinion of what is significant or not, just as I
> am entitled to my opinion of his opinion.

Have any open source projects written in Haskell ever garnered a significant 
number of users?

-- 
Dr Jon D Harrop, Flying Frog Consultancy Ltd.
http://www.ffconsultancy.com/products/?e