Drastic Prelude changes imminent

[Edward Kmett]

We started work on the Foldable/Traversable issue during 7.8, right after
the BBP process burned down and it was clear there was a strong desire to
see motion on that topic, and the core libraries committee was formed, but
decided not to act on anything during 7.8, as 7.8's release seemed immanent
at the time. Little did we know it'd drag on for another 7 months. Overall,
that proved to be a good thing, as we spent much of that time trying to
understand how to retie the Gordian knot that is the set of hs boot files
that makes up base.

But there is definitely a process concern. We've now had proposals that
'passed' a year ago or six months ago, that we're able now to implement,
now that we finally have folks in place that know how to work through the
existing process.

Most of these are small fixes, broadly agreed upon. Do we make someone who
already jumped through all the hoops to get a coherent proposal through the
process run and jump through them all again?

Once 7.10 started we set to work in earnest. A large part of the committee
work that has happened in 7.10 has been un-log-jamming issues that have
been stuck in trac on the libraries@ process with or without a patch. Some
of the bugs that had made it into trac had languished for years.

Most of the complications in the current design came about because of
things that only became obvious once the framework was in place.

I do agree that it is quite difficult to keep abreast of all that is
happening in GHC.

Part of that is a consequence of having an order of magnitude more cooks in
the kitchen these days. At the same time you have core libraries work going
on, we have a bunch of independent work going on on the front of how to
deal with SIMD instructions, swizzling, folks who care deeply about the
fate of DPH, a ton of exciting work on the cabal side that centers around
how to deal with finally solving 'cabal hell', complications induced by
considering the impact of Kilpatrick and Yang's work on Backpack, work on
the RTS and garbage collector, etc.

There are a lot more people active now. There were something like 40+
people actively contributing code at the beginning of last year, and that
number can only have grown since. When it was just the two Simons and
whatever couple of graduate students happened to stumble in, things
proceeded at a much more gentle pace. But with Simon Marlow moving on to
facebook, a lot of folks decided to pitch in to pick up the slack. It is
easy to keep track of the head-space of a smaller pool of contributors.

A whole new toolchain has been built up around using phabricator for code
review that has greatly improved the quality of the patches going into GHC.
This has helped a great deal, but if you think the libraries@ process is
hard to keep up with, well, there have been several hundred phabricator
patch reviews since the process started last year, mostly dealing with the
minutiae of how to integrate pre-existing trac tickets raised by libraries@
proposals or flat out ghc bug reports. ;)

We absolutely do need to work out more effective ways to communicate what
is going on.

There will be growing pains, there will be surprises, but we are all
pulling towards a common goal, that of making GHC be the best compiler it
can be for Haskell, and making continuing to strive towards making Haskell
be the best version of itself that it can be.

-Edward

On Wed, Jan 28, 2015 at 3:09 PM, Ben Gamari <ben at smart-cactus.org> wrote:

> Joachim Breitner <mail at joachim-breitner.de> writes:
>
> > Hi,
> >
>
> snip
>
> >  * Proposal wiki pages, with motivation, examples, alternatives,
> >    have served us well. I’m not sure why that was skipped here.
> >    We should have them in the future.
> >
> Perhaps just as important is to maintain a Wiki page listing the
> proposals being considered by the CLC as well as the release milestones
> for adopted proposals. As someone who followed the BBP proposal fairly
> carefully in 2013, I was also a bit surprised when patches started
> showing up this cycle.
>
> Cheers,
>
> - Ben
>
> _______________________________________________
> Libraries mailing list
> Libraries at haskell.org
> http://www.haskell.org/mailman/listinfo/libraries
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.haskell.org/pipermail/libraries/attachments/20150128/50f6c422/attachment.html>