English version
Accueil     À propos     Téléchargement     Ressources     Contactez-nous    

Ce site est rarement mis à jour. Pour les informations les plus récentes, rendez-vous sur le nouveau site OCaml à l'adresse ocaml.org.

Browse thread
[ANN] Kaputt 1.0 alpha
[ Home ] [ Index: by date | by threads ]
[ Search: ]

[ Message by date: previous | next ] [ Message in thread: previous | next ] [ Thread: previous | next ]
Date: 2008-11-24 (19:13)
From: forum@x9c.fr <forum@x...>
Subject: Re: [Caml-list] [ANN] Kaputt 1.0 alpha

Le 24 nov. 08 à 17:52, Yoann Padioleau a écrit :

> forum@x9c.fr writes:
> Hi Xavier,
>> Selon Stefano Zacchiroli <zack@upsilon.cc>:
>>> (...)
>> I was indeed aware of oUnit when I started working on Kaputt.
>> The initial intent was to provide OCaml with something along
>> the lines of the QuickCheck library developed for Haskell
>> (http://www.cs.chalmers.se/~rjmh/QuickCheck/). Clones of QuickCheck
>> exist for various languages (caml-list@inria.fr) but not for OCaml,
> Actually I posted a few years ago a small module imitating parts of  
> quicheck:
> http://aryx.kicks-ass.org/~pad/ocaml/quickcheck.ml

Sorry, I failed to remember this.

> Apparently they even now use my ugly file in the Janet Street Core  
> Library.
>> even if a QuickCheck equivalent is provided by the Reins library
>> (http://ocaml-reins.sourceforge.net/).
> I didn't find much in this ocaml reins library ... and it
> looks quite heavy in functors ...

Yes, I indeed wanted to rely upon functions/combinators instead.
It seems lighter this way.

> Your library looks very good, very clean. I just noticed you missed
> the 'pair' and 'tuple' random generators.

Well, maybe their names are not intuitive but Generator.zip{1..5}
allows to combine n generators into a generator returning tuples
of order n (zip1 being the identity function).

> One criticism is that like many other libraries, such as oUnit,
> ocamlcalendar, they seem a little bit heavyweight. You got hundreds
> of functions, 5 modules ...

That's true but it seems hard to have a combinator-based library with
few functions/combinators. With too few combinators, you leave too
much work to the developer (including a lot a boilerplate code).
With too many combinators, you request the developer to take more
time in order to use the library. I find it hard to calibrate the size  
the library.

Maybe things could be done in a lighter way by using objects ?