com.blockether.vis.ext.language-clojure.nrepl-client — com.blockether/vis-language-clojure 0.1.10

Thin, observable nREPL client for clj/eval.

Connection model:

• One nrepl.core/connect socket per [host port] key, cached on a defonce atom so we survive (require :reload) during development.
• ONE long-lived nREPL session per connection, cloned lazily on first use and cached beside the socket — then REUSED by every eval!. This is how Cider/Calva/every editor drives nREPL: no per-eval clone/close round-trip on the hot path (those were what blew the native-tool timeout budget under JVM load), nothing to leak, and session-local state — *1/*2/*3/*e and dynamic set!s — PERSISTS across calls like a real REPL ((def …) was already global; now the whole session is).
• Stale / closed sockets are detected (IOException / nil message stream) and the entry is evicted — closing the socket and dropping the cached session — so the next call re-dials + re-clones.

Returned shape (success) — STRING keys (crosses the strings-only boundary as a tool :result; enrichment adds "error_message"/"error_data"/"trace"): {"value" "42" ; pr-str of the LAST form's value, or nil "values" ["1" "42"] ; pr-str of every emitted value "out" "hello\n" ; stdout aggregated "err" "" ; stderr aggregated "ns" "user" ; final ns name "status" #{"done"} ; nREPL status set (strings) "ex" nil ; exception class name, when status :ex "root_ex" nil ; root exception class name "ms" 12 ; wall-clock duration "port" 7888 "timed_out" false}

Failure paths throw ex-info with :type :clj/nrepl-* so the Vis tool wrapper can surface a clean error to the model.

close-all! clj

(close-all!)

Close every cached connection. Idempotent. Useful from tests / doctor / shutdown hooks.

eval! clj

(eval! {:keys [host port code ns timeout-ms pretty? print-margin]
        :or {host "localhost" timeout-ms 30000 print-margin 100}})

Evaluate code in the nREPL at host:port. Opts: :host defaults to "localhost" :ns starting namespace, e.g. "user" :timeout-ms default 30000 :pretty? when true, ask nREPL's print middleware to pretty-print the value(s) SERVER-SIDE via nrepl.util.print/pprint — so the live object is formatted where it lives (handles unreadable objects / lazy seqs) and :value/:values come back as multi-line, indented text. Output stays valid EDN. :print-margin right-margin columns for pretty printing (default 100)

Always returns a map (see ns docstring). Connection failures throw :clj/nrepl-connect-failed; everything else (eval error, timeout) is reported inside the returned map so the model can read it as data.

evict! clj

(evict! host port)

Drop the cached connection (and its long-lived session) for [host port], CLOSING the socket so BOTH sides let go: the server reaps the session's executor thread and OUR client transport thread dies instead of parking forever. Call it whenever a REPL at [host port] goes away (stop, reap, crash) — a cache entry outliving its process leaks one thread + one socket per REPL. The next call re-dials and re-clones.

health-check! clj

(health-check! {:keys [host port timeout-ms]
                :or {host "localhost" timeout-ms 2000}})

REAL eval-based health check — the strong signal probe!'s describe can't give. Runs (+ 1 1) through a DEDICATED health session on the cached connection under a SHORT timeout and confirms it returns "2". describe only proves the accept loop answers; this proves the JVM actually EVALUATES code, so a wedged eval executor / blocked compile lock is caught even while describe still replies. Returns:

{:status :up :ms N} — evaluated "2" cleanly; the eval path is healthy. {:status :unresponsive :form "(+ 1 1)" :ms N :hint "..."} — connected but the eval TIMED OUT or didn't return "2": the REPL is wedged and should be killed & restarted (or reprobed). {:status :down :form "(+ 1 1)"} — could not connect (stale .nrepl-port / dead process).

Uses a SEPARATE session so it never clobbers the user session's *1/*2/*3. Never throws. On a hard socket error the connection is evicted so the next call reconnects fresh.

health-form clj

The canonical liveness eval: cheap, pure, and an unmistakable "2" result.

probe! clj

(probe! {:keys [host port timeout-ms] :or {host "localhost" timeout-ms 100}})

Best-effort liveness probe for the nREPL at host:port. Sends a single describe op (no code execution) under a SHORT timeout and classifies:

{:status :up :versions {:clojure ..} :dialect :clj|:cljs :cwd "/path"} — connected and the server answered describe. {:status :unresponsive} — socket opened but no clean describe reply in budget. {:status :down} — could not connect (stale .nrepl-port, dead proc).

:dialect (clj vs cljs) is read from the describe metadata; :cwd is the server JVM's working directory via one tiny eval (nil when unavailable). Both are best-effort and only present when :up.

Never throws. Reuses the cached connection (warming the same pool eval! uses). The short :timeout-ms (default 100) is passed to nrepl/client, which bounds each response read regardless of the cached connection's transport timeout.