[Haskell-cafe] Minimal Haskell Platform
Vincent Hanquez
tab at snarc.org
Wed Apr 16 08:58:11 UTC 2014
On 2014-04-15 09:45, Roman Cheplyaka wrote:
> * Simon Hengel <sol at typeful.net> [2014-04-15 13:02:38+0800]
>>> Would it be worthwhile to create a "Minimal Haskell Platform" to
>>> create a truly common platform that everybody would be happy using?
>> I think most Haskell developers use a reasonably recent version of GHC
>> (that would currently still be 7.6.3 for me) and the latest version of
>> cabal-install. Everything else can be installed with cabal-install.
>>
>> So yes, I think we should just have an easy way to get that minimal
>> setup (+ most importantly don't recommend something to beginners that we
>> don't use ourself).
>>
>> Where would something like the HP actually make sense? For stuff that
>> has external dependencies that are not easily installable with
>> cabal-install (like curses bindings, SSL support, etc.). We have none
>> of this in the HP. So I think currently we just have additional costs,
>> but no benefits (+ we harm innovation by arbitrarily "endorsing" random
>> packages).
> Agreed.
>
> As Tillmann notes, HP is indispensable on Windows, but that is a side effect.
> Indeed, HP as a simple way to conveniently install the bits that are otherwise
> hard to install makes perfect sense.
>
> However, nowadays most attention goes not to that, but to choosing ordinary
> cabal packages to be bundled with HP. I see little value in that.
>
> As a way to ensure the "installability" of packages, Stackage now does a far
> greater job than HP by:
>
> - having much wider selection of packages, and making it almost trivial to
> incorporate new ones;
> - constantly building packages and running tests;
> - notifying maintainers about broken packages and packages with restrictive
> upper bounds.
>
> In other words, it ensures that Stackage packages are installable at any time
> with cabal. (Well, strictly speaking, there may be a delay between when a
> problem is reported by Michael and when it's fixed by the maintainer.)
>
> Apart from that, HP "arbitrarily endorses" its packages and forces them into
> stability mode. Whether it is a good thing or not is another argument and
> depends on one's perspective. I know that some people find it important (and
> I'm well aware of their reasoning).
>
> Personally, with my hats of
>
> * an open source libraries developer/maintainer
> * a personal Haskell user
> * a commercial Haskell user
>
> on, I don't care about HP. The only time when I remember about it is when I
> tell novices how they can "install Haskell". But again, this has nothing to do
> with the bundled cabal packages, only with the way of installing ghc/cabal/etc.
Agreed with Simon and Roman here,
The initial goals of the HP were good, but I think a smaller minimal
haskell HP/installer (really for windows) would make more sense now
considering the situation.
I think for something like the HP to be successful (according to its
goals), you would need to have full time staff developing it together in
one common place.
So currently, it doesn't really provides batteries, but merely a base++
with an extra set of "random" packages (opengl, cgi, ..).
And, at the same time, it doesn't provide that much more API stability
either (across multiple versions of the HP).
--
Vincent
More information about the Haskell-Cafe
mailing list