[Haskell-cafe] is it normal for docs to be pending for 24 hours?

[Michael Snoyman]

On Sun, Nov 13, 2016 at 3:35 PM, Saurabh Nanda <saurabhnanda at gmail.com>
wrote:

> Hackage and Stackage serve different purposes.  Hackage is all packages,
>> while Stackage is a curated set of packages.  (Therefore, Hackage is
>> essentially the "upstream" of Stackage.)  We shouldn't give up on Hackage.
>>
>
> I know I'm jumping into a very, very touchy topic, but isn't the following
> possible:
>
> * Stackage & Hackage combine -- even the .cabal & .stack file formats (if
> possible)
>

Let's not conflate two things. I assume you're talking about stack.yaml as
the .stack file format. This should be a completely separate discussion for
multiple reasons:

* That's about Stack vs cabal-install instead of Stackage vs Hackage
* It's completely necessary to have package-level vs project-level
configuration (even cabal-install has a separate project config format
separate from the .cabal format)


> * The combined entity supports curated packages (via LTS) AND a global
> free-for-all package list. I, as a user, get to decide how I want my
> packages to be resolved -- via a curated LTS or via the global package
> index.
>

This was discussed in ernest at ICFP in 2014, and the resulting proposal
was GPS Haskell. The idea was that Hackage would add support for curated
package sets. Personally, I didn't think this was necessary, and
cabal-install should have just learnt logic to get information from
stackage.org so that adding the functionality to Hackage wasn't a blocker
for getting curated package sets available to users.

In reality, the curated package set feature never got added to Hackage,
cabal-install never added curated package set support, GPS Haskell was
abandoned, and Stack and LTS Haskell were created instead.


> * The combined entity focuses on building a kickass, unified, piece of
> community infra, instead of dividing effort+resources and solving similar
> problems twice.
>
>
>
It's a wonderful idea.

Michael
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.haskell.org/pipermail/haskell-cafe/attachments/20161113/a3ab4f76/attachment.html>