[ghc-steering-committee] Please review #669: Initial categorization of language extensions

[Malte Ott]

Dear Committee,

On 2024-10-29 19:45, Adam Gundry wrote:
> Trevis Elser proposes a classification of (some) language extensions based
> on previous work in proposal #601 establishing categories:
> 
> https://github.com/ghc-proposals/ghc-proposals/pull/669
> 
> https://github.com/telser/ghc-proposals/blob/initial-extension-categorization/proposals/0000-extension-categorization.md

(Note that additionally to adding the new proposal the PR also contains a small diff to the previous proposal #601.)

This proposal rested for a while, but after some revisions at Zurihac I am
recommending to accept this proposal. The proposal is the logical next step after
the acceptance of #601. It’s not pretty to have a categorization but not use it
and I think the classification is very helpful to guide users and us.

In the proposal Trevis nominates

72 extensions as stable,
10 extensions as deprecated,
 5 extensions as legacy,
 8 extensions as experimental

and

42 remain unclassified.

This only affects documentation and our future decisions. Only observable
change in GHC will be that deprecated extensions consistently get a warning
via -Wdeprecated-flags.

The intention is to basically just describe the status quo with the
classification and defer all contentious extensions to be classified later.

My general tendency is that we should err in the direction of marking
extensions stable because at least for widely used extensions, even those where
we are not very happy with the design, marking them experimental is like giving
us a free pass to break our stability principle. Also marking an extension
stable is not the same as "recommended" as it is perfectly fine for an extension
to be stable but not part of a GHCXXXX language edition.

Note that "For an extension to be classified as Stable it must be considered
Stable when used in combination of all possible, non-mutually exclusive,
extensions that are already Stable." I admit that I cannot judge this with
certainty for all listed extensions. To be honest I learned e.g. of the
existence of NoTraditionalRecordSyntax from reading the list. While I have been
told that the categorization has been produced in close collaboration with GHC
devs I ask everyone to keep this in mind when reviewing the list.

Please review the overall proposal and the proposed categories. You can give
your approval conditional on moving some extensions to "unclassified". Others
and I have already done so, so from my point of view the list is good as is.

Best,
Malte