[ghc-steering-committee] GHC2023

[Joachim Breitner]

Hi,

Am Mittwoch, dem 23.11.2022 um 17:51 +0100 schrieb Joachim Breitner:
> encouraged by Arnaud reports about Rust editions, and also in light of
> the ExplicitNamespaces “bug” in GHC2021, I’m now starting a proposal to
> define GHC2023. 
> 
> Please see
> https://github.com/ghc-proposals/ghc-proposals/pull/559
> https://github.com/ghc-proposals/ghc-proposals/blob/joachim/ghc2023/proposals/0000-ghc2023.rst
>
> The proposal text succinctly tries to alleviate worries that this is
> “too soon”, and gives motivation for the nominated extensions.
> 
> At the moment it suggest to add ExplicitNamespaces and LambdaCase
> 
> I invite all committee members to 
>  * nominate more extensions to add
>  * bring arguments in favor of an extension
>  * bring arguments against an extension
> 
> You can either add them directly (via suggested edits) to the proposal
> text, or write them here and I’ll paraphrase them into that document,
> to try to summarize the discussion well.
> 
> Once that phase seems concluded, I’ll invite you to vote. Not sure yet
> how precisely, but we’ll vote in a way that one can also express that
> no GHC2023 should happen.

I have not seen much engagement from the committee on this PR, and I
sense a certain reluctance to have another GHC20xx this year. But I
don’t want us to deviate from the original accepted plan just because
of implicit assumptions, nor get bogged down by the more fundamental
(and important!) discussion about the role of extensions in general.

So let’s bring this to a vote – if the camp that prefers even rarer
releases is in the majority, the vote will show that. 

I’m still fairly convinced that we are doing the community a service if
we take the idea of GHC20xx editions as a continuous and predictable
stream of improvements serious. Even if the delta is kinda small.
Actually, I should say: _Especially_ if the delta is kinda small! This
takes the anxiety out of “anew GHC20xx has been released”, just like
the now bi-yearly GHC releases ideally don’t scare anyone anymore, as
it helps us get used to and gain practice with the process.

So let’s collect possibly more nominations and otherwise discuss with
the community on the PR for two more weeks (until roughly Sun, Jan 22).
If you’d like to get an extension on the ballot, or expand the
rationale for an extension, simply edit 
https://github.com/ghc-proposals/ghc-proposals/blob/joachim/ghc2023/proposals/0000-ghc2023.rst

Then I will ask a for a vote. The ballot will roughly look like this:

  For each of the following extension, should it become part of
  GHC2023? Please annotate with “strong yes”, “weak yes” or “no”.

  GotoStatement
  SemanticComments
  ImplementationByChatGTP

We’ll have GHC2023 if _at leas one_ extension has a majority of “strong
yes” vote, and it will contain those extensions that have a majority of
“yes” votes.

This allows you to distinguish between “worth having GHC2023 for” from
“not worth having GHC2023 for, but good enough to include if we have
GHC2023 anyways”. For example if you think there shouldn’t be a
GHC2023, you just never vote “strong yes”, but you can still have a say
what should be in it if it happens, by voting “weak yes” or “no”.


Cheers,
Joachim






-- 
Joachim Breitner
  mail at joachim-breitner.de
  http://www.joachim-breitner.de/