ghciwatch
ghciwatch copied to clipboard
ghci-based file watching recompiler for Haskell development
> My only ask would be if we can get feedback per file before a full load cycle. Like some sort of incremental write to the ghcid file. With `--errors...
Right now, the TUI keeps all the output its ever seen. For long-lived sessions, this will consume arbitrarily large amounts of memory! Set a reasonable upper bound, like 10,000 lines...
The `--tui` is cool and we need a new framework to test it. I think [asciinema](https://asciinema.org/) fits the bill. The basic idea is: * Let users customize the set of...
It would be nice to be able to *also* use `ghciwatch` as a `ghci` session interactively. From [SyncLinear.com](https://synclinear.com) | [DUX-1367](https://linear.app/mercury/issue/DUX-1367/run-haskell-code-by-typing-it-into-a-prompt)
Probably requires [DUX-1287](https://linear.app/mercury/issue/DUX-1287/server-mode) and a VS Code plugin [https://mercurytechnologies.slack.com/archives/C01QJRR7E23/p1701886757792739](https://mercurytechnologies.slack.com/archives/C01QJRR7E23/p1701886757792739) From [SyncLinear.com](https://synclinear.com) | [DUX-1654](https://linear.app/mercury/issue/DUX-1654/reload-status-in-vs-code)
> Is there a way of reversing the order of errors so that the ‘first’ error is printed last? When I have more than one screen of errors I think...
(Previously: scripted mode) We want `ghcid-ng` to be able to listen for events via TCP and ideally respond to them, too. This will make it feasible to implement integration tests....
It wouldn't be *too* hard to swap out the `ghci` backend for one that just does a standard `cabal build`. This might be useful! From [SyncLinear.com](https://synclinear.com) | [DUX-1403](https://linear.app/mercury/issue/DUX-1403/ghc-backend)
The JSON test harness is super useful. We should split it out into its own crate for others to use. From [SyncLinear.com](https://synclinear.com) | [DUX-1396](https://linear.app/mercury/issue/DUX-1396/split-out-test-harness)