wither the Platform

[Mark Lentczner]

A common sentiment I see here is that the Platform always seems out of
date. That has always been true, and to a degree by design: It is the
stable set, not the wild west. Historically, the first release of a major
new GHC was seen as needing a bit of a trial period before we enshrined it
in a Platform release. Hence, they lagged. There are also other concerns
like stable release schedule (which the Platform has been poor at, but has
always been a goal).

GHC has made great strides in being much more stable out of the gate. And
the Platform has been significantly automated recently. For example, there
is already an Alpha of the Platform for 7.10 RC3. But it still takes
person-power to run it, test it, and get it out. And there is still the
issue of stable release times.

If you want latest GHC, and latest libs - then Platform is not the way to
go. A solid question is: Where do we want to direct the mass of the
community? At head? At latest release? At some older, but now stabilized
point? What do we want their experience to be? We will best served if the
answer is pretty much right for the bulk of the community: not just
newcomers, not just people doing production work, not just hobbiests... all
of them.

Most other language systems manage to have a single distribution that works
for the majority of the community, and most of them are content to have
much better stability and support. Python put out 2.7.0 in 2010, and 2.7.9
last Dec. It is still supporting 2.7 line (with source and library
compatibility) while it is off releasing 3.3 line. This would be for us
like deciding to support and update 7.8, and the pre-FTP Prelude for
another four years. Which wouldn't be so bad...



​
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.haskell.org/pipermail/ghc-devs/attachments/20150323/05266e84/attachment.html>