feat(helix): add optional Steel plugin client - #94
Draft
icorbrey wants to merge 2 commits into
Draft
Conversation
Adds clients/helix-steel/ for people running a Steel-enabled Helix build (mattwparas/helix, branch steel-event-system — Steel is not yet in mainline Helix; helix-editor/helix#8675 is still open). cogs/badjuju-core.scm holds pure logic (window classification, JJ: shortcut-line detection, file:// URI stripping, cursor-arg construction) with no editor-context dependency, unit-tested directly against a plain `steel` interpreter. cogs/badjuju.scm wraps every badjuju.* server command as a named Steel command and adds: - Auto-open after state-changing commands. Vanilla Helix code actions discard the workspace/executeCommand JSON result (helix-view::execute_lsp_command only logs errors on it, by design — Helix expects state changes via workspace/applyEdit instead), which is why the plain languages.toml client documents manual `:open` as a known limitation. jj-execute! opens the returned URI itself. - RET context dispatch in log.jujutsu (apply a JJ: shortcut vs. goto-definition), matching the Neovim/Emacs clients — something plain Helix has no keybinding layer to express at all. - A `tab`-prefixed keymap submenu for magit-style single-letter access. Deliberately not a flat top-level scheme like Kakoune/ Neovim/Emacs ship: Steels extension/label keymap is a static override for every buffer sharing the .jujutsu extension, with no per-window-kind fallback to a native binding the way those clients per-buffer keying has. A flat letter scheme would silently break normal editing in describe.jujutsu and hunk-edit.jujutsu, which are real text-entry buffers. RET and tab both have no default Helix normal-mode binding, so nesting everything else under tab costs one keystroke instead of any collision risk. Testing gap this fills: the plain languages.toml Helix client has no Steel logic to test, so this is the first Scheme code in the repo. test/badjuju-test.scm unit-tests badjuju-core.scm directly. test/smoke.scm loads the real badjuju.scm against stub helix/* modules (test/fixtures/helix/) and calls every exported command across status/log/squash/describe window kinds — this caught three real bugs during development (a missing `helix/misc.scm` require for send-lsp-command, a missing `set-scratch-buffer-name!` stub surfacing the same gap, and `(void)` used as a call instead of the bare value `void`, which is not callable in Steel). No Steel-enabled Helix build exists in this repos CI, so neither suite proves in-editor behavior — thats documented explicitly in the README rather than implied by a passing test. CI: new "Helix (Steel)" job installs the standalone `steel` interpreter (steel-interpreter on crates.io, same 0.8.2 the tests were developed against) and hard-fails via `steel --version` if setup silently no-ops, mirroring the Neovim/Emacs jobs guard against run.sh soft-skipping when the tool is missing.
`new` ("Create a new scratch buffer") is a typed command from
helix/commands.scm (imported with the `helix.` prefix), not a static
command from helix/static.scm (imported with `helix.static.`).
`(helix.static.new)` is a FreeIdentifier that only surfaces when Steel
compiles the whole file at load time, not when jj-show-scratch! is
actually called -- which is why the standalone-steel smoke test in
test/smoke.scm never caught it: its fixtures/helix/static.scm stub
defined its own `new` to make the call succeed, masking that the real
static.scm has no such export.
Caught by driving a real Steel-enabled `hx` (mattwparas/helix,
steel-event-system-derived megamerge) through `expect` and capturing
the TUI error popup via a terminal emulator (pyte) -- the popup is the
only place present_error_inside_engine_context surfaces a load error;
it is never logged. Confirmed the fix by the presence of
"Finished loading init.scm!" in the log, which only prints on the
Ok(_) arm of the whole-file eval.
Testing gap this exposes: the smoke test stubs every helix/* module
by hand instead of validating call sites against real exports, so a
misattributed module for any helix.*/helix.static.* call is invisible
to it. Left as a known limitation in the plugin README rather than
fixed here -- doing so properly means either vendoring the real
static.scm/commands.scm export lists to diff against, or dropping the
hand-written stubs for a real Steel-enabled hx in CI, neither of which
is a quick follow-up.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Yo. I have not looked at a lick of this code nor tested it, probably not the best idea to run it yet.