I'm happy to accept this proposal. It does have a small downside, as I discussed with Oleg and Teo on GitHub, which is that it becomes harder for a library to define a Quasi instance not based on Q, e.g. for test code, as implementing qRunQ requires access to internals. But this is a very rare thing to want and arguably there are better alternatives, so I'm fine with the trade-off. Our discussion also explored the longer-term possibilities for making bigger changes to the template-haskell library API, e.g. potentially removing Quasi altogether (or at least the partial Quasi IO instance). There I think the right strategy is to accept this proposal so the interface can be decoupled from GHC's implementation, then explore interface changes independently. Cheers, Adam On 06/04/2026 19:11, Sebastian Graf wrote:
Dear committee,
Teo proposes to hide the implementation of the |Q| monad. This allows an internal only interface to be split out of |Quasi|. This enables the internal interface to evolve independently of |Quasi|, the external interface. This small change has a practically important benefit: adding a new function to Q (a seemingly minor change) no longer forces a major version bump of GHC to cause a major version bump of template-haskell that will transitively ripple through the whole ecosystem. I highly recommend we accept the proposal.
Sebastian
Am Di., 10. März 2026 um 10:55 Uhr schrieb Simon Marlow <marlowsd@gmail.com <mailto:marlowsd@gmail.com>>:
Teo Camarasu submitted proposal *#700: Abstract Q* to the committee for review.
https://github.com/ghc-proposals/ghc-proposals/pull/700 <https:// github.com/ghc-proposals/ghc-proposals/pull/700>
I'd like to nominate *Sebastian Graf *as the shepherd.
Please guide us to a conclusion as outlined in https://github.com/ghc-proposals/ghc-proposals/blob/master/ README.rst#committee-process-for-responding-to-a-proposal <https:// github.com/ghc-proposals/ghc-proposals/blob/master/ README.rst#committee-process-for-responding-to-a-proposal>. If possible, please aim to make a recommendation by *24 March 2026*.
Cheers Simon -- Adam Gundry, Haskell Consultant Well-Typed LLP, https://www.well-typed.com/
Registered in England & Wales, OC335890 27 Old Gloucester Street, London WC1N 3AX, England