nomarmy 0.1.0-alpha.2 → 0.1.0-alpha.21

package/scripts/lib.sh

@@ -19,9 +19,13 @@ nomarmy_compat_prefix(){
   done
 }
 
+# Your settings live here, over the package's config/ defaults, so an update
+# can't reset them (lib/user-config.mjs does the same for the CLI).
+nomarmy_user_config_dir(){ echo "${NOMARMY_CONFIG_DIR:-${XDG_CONFIG_HOME:-$HOME/.config}/nomarmy}"; }
+
 nomarmy_available_profiles(){
   local root; root="$(nomarmy_root)"
-  find "$root/config/profiles" -name '*.env' -exec basename {} .env \; | sort | tr '\n' ' '
+  { find "$root/config/profiles" -name '*.env' -exec basename {} .env \; ; find "$(nomarmy_user_config_dir)/profiles" -name '*.env' -exec basename {} .env \; 2>/dev/null; } | sort -u | tr '\n' ' '
 }
 
 # Where this install's models run (lib/execution.mjs is the same logic for the
@@ -129,7 +133,8 @@ load_profile(){
     elif command -v nvidia-smi >/dev/null 2>&1; then profile=nvidia-linux
     else profile=cpu-linux; fi
   fi
-  profile_file="$root/config/profiles/$profile.env"
+  profile_file="$(nomarmy_user_config_dir)/profiles/$profile.env"
+  [[ -f "$profile_file" ]] || profile_file="$root/config/profiles/$profile.env"
   if [[ ! -f "$profile_file" ]]; then
     echo "ERROR: unknown profile '$profile'. Available: $(nomarmy_available_profiles)" >&2
     exit 2
@@ -137,6 +142,8 @@ load_profile(){
   set -a
   # shellcheck disable=SC1091
   source "$root/config/common.env"
+  # shellcheck disable=SC1091
+  [[ -f "$(nomarmy_user_config_dir)/common.env" ]] && source "$(nomarmy_user_config_dir)/common.env"
   # shellcheck disable=SC1090
   source "$profile_file"
   set +a

package/scripts/select-model.mjs

@@ -3,12 +3,17 @@ import { createInterface } from "node:readline/promises";
 import { stdin as input, stdout as output } from "node:process";
 import { readFile, writeFile } from "node:fs/promises";
 import path from "node:path";
+import { userCommonPath } from "../lib/user-config.mjs";
+import { fileURLToPath } from "node:url";
 
-const root = path.resolve(path.dirname(new URL(import.meta.url).pathname), "..");
+// fileURLToPath, not URL.pathname: on Windows that gave "/C:/Users/Jason%20Pugh/...",
+// a stray slash and an encoded space, and connect failed copying from C:\\C:\\...
+const root = path.resolve(path.dirname(fileURLToPath(import.meta.url)), "..");
 const args = process.argv.slice(2);
 let query = "";
 let repo = "";
-let configPath = path.join(root, "config", "common.env");
+// Your settings file, not the package's config/ (an update would reset it): lib/user-config.mjs.
+let configPath = userCommonPath();
 
 for (let index = 0; index < args.length; index += 1) {
   const value = args[index];
@@ -70,16 +75,18 @@ try {
   const alias = (await rl.question(`Local model alias [${defaultAlias}]: `)).trim() || defaultAlias;
   if (!/^[A-Za-z0-9._-]+$/.test(alias)) throw new Error("Alias may contain only letters, numbers, dots, underscores, and hyphens.");
 
-  console.log(`\nReady to update ${path.relative(root, configPath)}:`);
+  console.log(`\nReady to update ${configPath}:`);
   console.log(`  NOMARMY_MODEL_REPO=${repo}`);
   console.log(`  NOMARMY_MODEL_QUANT=${requestedQuant}`);
   console.log(`  NOMARMY_MODEL_ALIAS=${alias}`);
-  if ((await rl.question("Apply this model configuration? [y/N] ")).trim().toLowerCase() !== "y") {
+  if (["n", "no"].includes((await rl.question("Apply this model configuration? [Y/n] ")).trim().toLowerCase())) {
     console.log("Canceled; no model configuration changed.");
     process.exit(0);
   }
 
-  const existing = await readFile(configPath, "utf8");
+  // Your settings file may not exist yet: start it rather than fail.
+  const existing = await readFile(configPath, "utf8").catch(() => "");
+  await import("node:fs").then((fs) => fs.mkdirSync(path.dirname(configPath), { recursive: true }));
   const replace = (key, value, text) => {
     const line = `${key}=${value}`;
     return new RegExp(`^${key}=.*$`, "m").test(text) ? text.replace(new RegExp(`^${key}=.*$`, "m"), line) : `${text.trimEnd()}\n${line}\n`;

package/scripts/start-inference.sh

@@ -15,8 +15,8 @@ if [[ -f "$PID" ]] && kill -0 "$(cat "$PID")" 2>/dev/null; then echo "llama-serv
 BIN="$NOMARMY_INSTALL_ROOT/llama-server"; [[ -x "$BIN" ]] || { echo "ERROR: $BIN missing"; exit 1; }
 ARGS=(-hf "$NOMARMY_MODEL_REPO:$NOMARMY_MODEL_QUANT" --alias "$NOMARMY_MODEL_ALIAS" --host "$NOMARMY_LLAMA_HOST" --port "$NOMARMY_LLAMA_PORT" -c "$NOMARMY_LLAMA_CONTEXT" -np "$NOMARMY_LLAMA_PARALLEL" -ngl "$NOMARMY_LLAMA_GPU_LAYERS" -t "$NOMARMY_LLAMA_THREADS" --metrics)
 # Everything below is optional advanced tuning, unset by default: nothing
-# here changes behavior for anyone who hasn't set these. See README's
-# "Advanced llama-server tuning" section for what each one is for.
+# here changes behavior for anyone who hasn't set these. See "Advanced
+# llama-server tuning" in docs/configuration.md for what each one is for.
 [[ -n "${NOMARMY_LLAMA_CACHE_TYPE_K:-}" ]] && ARGS+=(--cache-type-k "$NOMARMY_LLAMA_CACHE_TYPE_K")
 [[ -n "${NOMARMY_LLAMA_CACHE_TYPE_V:-}" ]] && ARGS+=(--cache-type-v "$NOMARMY_LLAMA_CACHE_TYPE_V")
 [[ -n "${NOMARMY_LLAMA_FLASH_ATTN:-}" ]] && ARGS+=(--flash-attn "$NOMARMY_LLAMA_FLASH_ATTN")