wither the Platform
eir at cis.upenn.edu
Mon Mar 23 01:35:45 UTC 2015
I have to say that I'm quite surprised this conversation is happening at all. As far as I knew before doing some research in advance of this post, the Haskell Platform is *the* way to install Haskell on a fresh machine. It's certainly what I've relied on in getting Haskell on my machines (both MacOS 10.8). While I personally don't feel the need for the HP to have some curated set of libraries (especially now that there are other curated sets available), I do feel a strong need to have it be nice, shiny, and easy to install. At least for Mac, I don't know of a suitable replacement.
I feel like we, as a community, ask a tremendous amount from our users.* By eliminating the HP, we'll be asking more from users before they're even a proper part of community. This seems like a step in the wrong direction, to me.
* Here are a few ways in which we ask a ton from users:
- Users have to figure out which libraries to use. Many basic tasks (e.g. parsing, regular expressions) have competing packages, and it's hard to know which is appropriate.
- Users have to deal with a very intricate type system. Basic libraries (like the new Prelude, `vector`, `lens`) use this complexity to their advantage, but perhaps to newcomers' disadvantage. (I'm well aware I'm, in some degree, to blame here!)
- Users have to deal with long, intricate error messages. This comes hand-in-hand with the previous point. We GHC hackers try our best, but I know we fall short of the mark here.
- Once a year, when the new GHC comes out, everything breaks.
- Cabal hell.
In return, the community gives and gives and is ever patient with newcomers. This is wonderful, and I attribute Haskell's growth to the friendliness of the community. But we should be aware of just how steep the curve is.
On Mar 22, 2015, at 12:08 PM, Joachim Breitner <mail at joachim-breitner.de> wrote:
> Am Sonntag, den 22.03.2015, 10:52 +0100 schrieb Herbert Valerio Riedel:
>> Currently GHC/Cabal knows about a global package db and a user package
>> db (the user pkg db is is what gets replaced/shadowed by cabal
>> sandboxes). Maybe we need a 3rd package db sitting between the global
>> and the user package db that interacts better with cabal sandboxes?
> this would also be great for distributions, which also provide packages
> that should not (necessarily) in sandboxes, but are installed system
> Joachim “nomeata” Breitner
> mail at joachim-breitner.de • http://www.joachim-breitner.de/
> Jabber: nomeata at joachim-breitner.de • GPG-Key: 0xF0FBF51F
> Debian Developer: nomeata at debian.org
> ghc-devs mailing list
> ghc-devs at haskell.org
More information about the ghc-devs