[Haskell-community] Request for comment: New haskell.org download page

[Jason Dagit]

[Sorry all about my previous email, I accidentally clicked "Send" before
typing anything.]

I don't personally get value from the HP, but my understanding is that
universities do. In particular, I've heard stories of professors teaching
Haskell courses and being able to ask their admin staff to "Install the
Haskell Platform" on the lab machines and then they tailor the course
materials to content in the HP.

If the HP were to go away, I feel strongly that we should have a
deprecation phase. Otherwise the users who do use it will feel like the rug
was pulled out from under them. Let's not cause that sort of pain if we
have the power to avoid it.

I agree with Gershom that having a minimal installer is the right way to
get going for professionals and hackers. I'm less in tune with beginners
these days, but I think Gershom is basically right when he says the are a
very diverse group and there isn't just one profile that fits the majority
of them.

On a personal side, I'm not sure what is entailed in giving objective info
about stack vs. cabal. I'm probably OK with it for some definitions, but if
it means giving stack equal status, then I feel that is a bit premature at
this point. In particular, cabal has the advantage of being established,
mature, and battle hardened. Stack, in my opinion, still needs time to
prove itself. Both have flaws, both have advantages, but only cabal has a
track record at the moment. Yes, I'm being conservative here, but I think
that's reasonable if we're talking about making recommendations to such a
wide audience.


On Thu, Sep 24, 2015 at 10:54 AM, Richard Eisenberg <eir at cis.upenn.edu>
wrote:

>
> On Sep 24, 2015, at 1:38 PM, Bryan Richter <b at chreekat.net> wrote:
> >  Since I don't know anyone who likes or uses
> > the HP, I think hearing their perspective would be valuable.
>
> At the threat of being booed out of the community, I'll stand up and say I
> like the HP. Of course it should improve -- and I like the direction it's
> going -- but I've used it as it is and would do so again.
>
> I think the observation that the HP promotes a ghc-centric view is why I
> like it. When working with a new language, I really do just want to think
> about the compiler. Only when I get around to trying to actually produce
> software do I care about a build tool. Starting with stack or cabal,
> instead of with ghc, means a much bigger cognitive load right away. Not
> only to you have Haskell files, but you also have .cabal files, and perhaps
> stack.yaml files. You have to think about directory structure -- both
> `cabal` and `stack` behave differently when they're run in or under a
> directory with project files than elsewhere.
>
> But, when I'm starting out, and when my students are starting out, I just
> want to think about the language. Compilers have, at their core, a simple
> interface: program text in, program binary out. Simple! It may be totally
> insufficient for producing portable libraries and distributable software,
> but I'm not tackling those issues on my first day. The fact that the HP
> means I can install something and right away tinker in GHCi and explore
> some common libraries is great.
>
> I'm not trying to derail the conversation. Or even to urge strongly that
> HP remain on the downloads page. I've not followed the thread closely
> enough to be able to make such a stand. But while scrolling through, I saw
> Bryan's request, and I thought I'd answer it.
>
> Happy booing! :)
>
> Richard
> _______________________________________________
> Haskell-community mailing list
> Haskell-community at haskell.org
> http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-community
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.haskell.org/pipermail/haskell-community/attachments/20150924/21df7db0/attachment.html>