Drop support for ghc-8.10

[jorisdral]

Recently, cardano-node releases have started using ghc-9.6 instead of ghc-8.10. Dropping support for ghc-8.10 in lsm-tree makes our lives a bit easier, because we can use less conditional compilation in our codebase. We'll now support versions ghc-9.2 and up, because we weren't explicitly supporting ghc-9.0 anyway

[wenkokke]

If it's all there, shouldn't we just do a first release with support for 8.10 and drop it in the second?

[jorisdral]

If it's all there, shouldn't we just do a first release with support for 8.10 and drop it in the second?

I don't think it's very useful to do so. I can understand it from the point of view that there is at least some released version that supports ghc-8.10 for users that want it, but we won't commit to supporting it into the future. Not supporting it from the get-go is in the end more useful for us as developers (no future support, no backporting, less maintenance, no users asking for backports/support). Morever, users that really want to use ghc-8.10 wouldn't be able to meaningfully upgrade their dependency on lsm-tree anyway if we drop it immediately in the next release

[jeltsch]

Looks pretty good to me. Just one comment further below and the following two recommendations:

• The following code should be removed:

  lsm-tree/cabal.project.release

  Lines 25 to 29 in 97f498a

  	-- TODO: fix assembly error in text-2.1.2
  	if (impl(ghc < 9.0) && os(windows))
  	package text
  	flags: -simdutf

• In the following code, the lower bound for the base version should be changed to 4.16:

  lsm-tree/scripts/generate-readme.hs

  Line 4 in 97f498a

  , base >=4.12

[dcoutts]

LGTM!