fix: disk picker fd0 (#112), menu Back guard (#106), offline theme (#113)

- #112: list_installable_disks drops fd/loop/sr/zram/<8GiB, largest-first so OVMF no longer offers /dev/fd0 as the default wipe target - #106: tools/check-menu-back.py + checks.menu-back (proved to fail) - #113: offline theme-switch contract documented (default/pinned only); theme-sync run_switch adds offline-oriented hint when no network Verified: V0 flake check; installer-safety; menu-back; py_compile.
2026-07-15 10:02:19 +01:00
parent 748d4af414
commit e6be5a5770
10 changed files with 202 additions and 49 deletions
--- a/agent/BACKLOG.md
+++ b/agent/BACKLOG.md
@@ -72,28 +72,6 @@ buttons in the installer (VM gum screens under Boreal look legible —
 artifacts in `/tmp/nomarchy-v2-swap-93/`). Needs Bernardo's screenshot or
 an Acer repro to pin the actual widget before fixing.

-### 113. Live session: offline theme switch rebuilds the world from source
-
-Surfaced by #99's evidence run (2026-07-13, `/tmp/nomarchy-v2-theme-99/`):
-in the **live ISO** session, offline `nomarchy-theme-sync apply gruvbox`
-fails — its `home-manager switch` tries to build **1176 derivations** from
-the `stage0`/`hex0` bootstrap up, then dies on offline source fetches
-(`vala`, `yasm`, `config.jsonc`, `ghostty-config`, `easyeffects.svg`,
-`typogrify`…), ending `error: Build failed due to failed dependency`. This
-contradicts docs/TESTING.md item 6 ("apply gruvbox → switch runs offline").
-The failing generation is `bvl30ggn…-home-manager-generation.drv`; it is
-**not** the drv the ISO pins (the pin is the *default*-theme
-`homeConfigurations.${username}.activationPackage`, i.e. Boreal — confirmed
-drv-identical to the failing one only for gruvbox state, so a non-default
-theme's generation is unpinned). NOT caused by #99's git-seed: the drv is
-byte-identical path-seeded vs git-seeded. **Decide the contract:** either
-pin every baked theme's live activationPackage into the ISO (size cost) or
-scope the offline-switch promise to the shipped theme + correct
-TESTING.md/MEMORY. Pass = a fresh live ISO either switches to any baked
-theme offline, or fails with a clear "needs network for this theme" message
-and accurate docs. **Not caused by anything in today's batch** — separate
-pre-existing ISO-pinning gap.
-
 ## NEXT

 ### 122. A crisper Nomarchy logo in fastfetch
@@ -283,14 +261,6 @@ scoped implementation/verification tasks before work starts. Pass = every
 current route has one deliberate home, navigation remains shallow, and no
 root-level entry is added or lost.

-### 106. Internal menu Back/Left navigation contract
-
-Reproduce the reported internal leaf with neither Back nor Left behavior, then
-fix any breach of the already-intended navigation contract. Add a permanent
-guard covering every internal leaf; external GUIs and free-text prompts retain
-their explicit Esc exceptions. Pass = no internal leaf can strand the user and
-the guard fails on a regression.
-
 ### 108. Keybindings menu presentation and completeness audit

 Group the menu as Window, Workspace, Menu, and Media, then prove every live
@@ -328,16 +298,6 @@ generation, System boot generation, and Files/root BTRFS scopes. Pass = labels
 state what is restored and from where before action, with destructive or reboot
 effects made clear and each existing recovery path represented once.

-### 112. Installer disk-picker safety
-
-Exclude floppy, pseudo, tiny, and otherwise non-installable devices such as
-`/dev/fd0`; never select them by default. **Reproduced in QEMU 2026-07-13**
-(#93's V2 run): with OVMF the guest exposes `/dev/fd0` and the picker listed
-it FIRST — a blind Enter selected it (artifacts:
-`/tmp/nomarchy-v2-swap-93/20-review.png`, `Disk: /dev/fd0 (WILL BE ERASED)`). Pass = the real install disk is clear,
-invalid devices cannot reach destructive setup, a permanent guard covers the
-filter/default logic, and a pre-destructive KVM run proves the picker behavior.
-
 ## LATER

 - **Wallpapers artifact split** (ROADMAP § Faster switches — decided,
--- a/agent/JOURNAL.md
+++ b/agent/JOURNAL.md
@@ -19,6 +19,19 @@ Template:

 ---

+## 2026-07-15 — Fix batch #112 disk picker, #106 menu Back, #113 offline theme
+- **Task:** Second fix-only batch from NEXT/NOW residuals.
+- **Did:** (112) `list_installable_disks` excludes fd/loop/sr/zram/tiny
+  (&lt;8 GiB), sorts largest-first; installer-safety markers. (106)
+  `tools/check-menu-back.py` + `checks.menu-back` (proved to fail when Back
+  stripped). (113) offline theme contract: TESTING item 6 scoped to default/
+  current theme; `run_switch` prints clear offline-oriented message on
+  network/build failures.
+- **Verified:** V0 flake check; installer-safety; menu-back prove-fail;
+  py_compile theme-sync.
+- **Pending:** V2 unattended install with OVMF fd0 present (optional).
+- **Next suggestion:** #108 keybinds audit, or #94 test-install chromium.
+
 ## 2026-07-15 — Fix batch #123–#126 (install bake, flake ref, ISO name, display)
 - **Task:** Fix-only NOW batch: #123 dconf pre-activate, #124 main flake
  seed, #125 ISO filename, #126 display mode list.
--- a/agent/MEMORY.md
+++ b/agent/MEMORY.md
@@ -137,6 +137,11 @@ iteration would otherwise rediscover.
  without a hierarchy pass.

 ## Gotchas (cost a debugging session once)
+- **OVMF exposes /dev/fd0 as TYPE=disk (#112):** disk picker must exclude
+  `/dev/fd*` and tiny sizes; sort largest-first or blind Enter picks floppy.
+- **Live ISO offline theme switch (#113):** only the *default/pinned* HM
+  generation is offline-safe; other presets may try to build the world —
+  document the contract and fail with a network-oriented message.
 - **Installer HM pre-activate needs XDG_RUNTIME_DIR (+ session bus) (#123):**
  `runuser … activate` inside `nixos-enter` has no user session. Without
  `mkdir -p /run/user/$UID` owned by the install user and
--- a/docs/ROADMAP.md
+++ b/docs/ROADMAP.md
@@ -404,6 +404,11 @@ Design/decision records and a running log of shipped work (items marked
  decision rather than a drive-by. Proved to fail: dropping snapshot makes it
  name the missing entry. **V3 pending** — that the apps *launch* needs real
  hardware (HARDWARE-QUEUE, Acer M5-481T).
+- ✓ **Fix batch #112 / #106 / #113:** installer disk picker drops fd0/loop/
+  optical/&lt;8 GiB and lists largest-first (OVMF no longer defaults to floppy);
+  `checks.menu-back` guards every internal rofi list emits ↩ Back; offline
+  theme-switch promise scoped to the live-pinned (default) theme with a clear
+  error when a non-pinned preset needs network.
 - ✓ **Install fix batch (#123–#126):** (123) HM pre-activate creates
  `/run/user/$UID` + `dbus-run-session` so dconf no longer aborts the bake;
  (124) main-built ISOs seed `?ref=main` and installer actually rewrites
--- a/docs/TESTING.md
+++ b/docs/TESTING.md
@@ -56,8 +56,11 @@ session may not start.
  Hyprland via `initial_session`; logging out lands on tuigreet.
 - The flake is seeded writable at `~/.nomarchy` (from the read-only
  `/etc/nomarchy` copy) — `$NOMARCHY_PATH` already points there.
- Locked flake inputs are pinned into the ISO store, so
-  `home-manager switch` works **without a network**.
+- Locked flake inputs are pinned into the ISO store, so a
+  `home-manager switch` for the **already-active / default theme** works
+  **without a network**. Switching to an arbitrary other preset can still
+  need downloads (that preset's HM generation is not in the pin) — the
+  tool should fail with an offline-oriented message (#113).

 ## 3. Verification checklist

@@ -70,7 +73,7 @@ Work through these in order; each one exercises a different layer.
 | 3 | `SUPER+Return` opens Kitty with the default theme colors (Boreal) | terminal default, ANSI palette |
 | 4 | `btop` in the terminal is themed | per-theme asset baking |
 | 5 | `nomarchy-theme-sync list` prints 24 presets | package, baked themes dir |
-| 6 | `nomarchy-theme-sync apply gruvbox` → state written, `home-manager switch` runs, desktop re-themes, wallpaper changes | the whole engine: state write, pure eval, HM rebuild, wallpaper hook |
+| 6 | `nomarchy-theme-sync apply boreal` (or re-apply the **current** theme) offline → switch succeeds. Applying a *different* preset may need a network — the ISO pins the default theme's generation, not every preset (#113) | state write + offline rebuild contract |
 | 7 | `SUPER+SHIFT+T` cycles wallpapers instantly (try `tokyo-night` for 4, or `boreal` for its set) | the runtime wallpaper path |
 | 8 | `nomarchy-theme-sync apply summer-night` → after the switch the bar has its own identity (light bar, different styling) | whole-swap waybar.css assets |
 | 9 | Open a GTK app — dark theme matching the palette | Stylix layer |
--- a/flake.nix
+++ b/flake.nix
@@ -456,7 +456,8 @@

          # Installer safety guards (item 54): offline compose-lock fail-closed
          # (no network flake lock), disk-signature warn regex, pre-activate
-          # recovery hint path, password floors. Pure source+regex contract.
+          # recovery hint path, password floors, disk-picker filter (#112).
+          # Pure source+regex contract.
          installer-safety = pkgs.runCommand "nomarchy-installer-safety"
            { nativeBuildInputs = [ pkgs.python3 ]; }
            ''
@@ -467,6 +468,15 @@
              touch $out
            '';

+          # Menu Back/Left contract (#106): every internal rofi_menu list
+          # emits ↩ Back (external modi / free-text keep Esc exceptions).
+          menu-back = pkgs.runCommand "nomarchy-menu-back"
+            { nativeBuildInputs = [ pkgs.python3 ]; }
+            ''
+              python3 ${./tools/check-menu-back.py} ${./modules/home/rofi.nix}
+              touch $out
+            '';
+
          # Installer keyboard no-variant regression (item 91): gum's `(none)`
          # row is display-only.  Exercise it through the real template
          # patcher, then build the exact us + empty-variant console keymap
--- a/pkgs/nomarchy-install/nomarchy-install.sh
+++ b/pkgs/nomarchy-install/nomarchy-install.sh
@@ -224,13 +224,43 @@ live_disk=""
 live_src=$(findmnt -no SOURCE /iso 2>/dev/null || true)
 [[ -n "$live_src" ]] && live_disk=$(lsblk -no PKNAME "$live_src" 2>/dev/null || true)

-mapfile -t disks < <(lsblk -dpno NAME,SIZE,MODEL,TYPE \
-    | awk -v skip="/dev/${live_disk:-NONE}" \
-        '$NF == "disk" && $1 != skip && $1 !~ /loop|zram|sr[0-9]/ {NF--; print}')
-[[ ${#disks[@]} -gt 0 ]] || fail "No installable disks found."
+# Installable whole disks only (#112): drop floppy/pseudo/tiny devices that
+# OVMF and some firmwares expose as TYPE=disk (e.g. /dev/fd0 first in the
+# list — a blind Enter would erase nothing useful and break the install).
+# Bytes via lsblk -b; 8 GiB floor catches fd0/USB crumbs; REQUIREMENTS still
+# recommend ≥40 GiB for a real install.
+# NOMARCHY_MIN_DISK_BYTES overrides the floor (tests / expert installs).
+MIN_DISK_BYTES="${NOMARCHY_MIN_DISK_BYTES:-$((8 * 1024 * 1024 * 1024))}"
+list_installable_disks() {
+    local skip="/dev/${live_disk:-NONE}"
+    local name size type model hsize
+    # NAME SIZE TYPE in bytes (-b); MODEL looked up separately (may contain spaces).
+    {
+        while read -r name size type; do
+            [[ "$type" == "disk" ]] || continue
+            [[ "$name" == "$skip" ]] && continue
+            case "$name" in
+                /dev/loop*|/dev/zram*|/dev/sr*|/dev/fd*|/dev/ram*|/dev/nbd*|/dev/md*)
+                    continue ;;
+            esac
+            [[ "$size" =~ ^[0-9]+$ ]] || continue
+            (( size >= MIN_DISK_BYTES )) || continue
+            hsize=$(lsblk -dno SIZE "$name" 2>/dev/null || echo "?")
+            model=$(lsblk -dno MODEL "$name" 2>/dev/null | tr -s '[:space:]' ' ' | sed 's/^ //;s/ $//')
+            printf '%s %s %s\n' "$name" "$hsize" "${model:-}"
+        done < <(lsblk -dpbno NAME,SIZE,TYPE 2>/dev/null)
+    } | sort -k2 -h -r   # largest first — never default to a tiny leftover
+}
+
+mapfile -t disks < <(list_installable_disks)
+[[ ${#disks[@]} -gt 0 ]] || fail "No installable disks found (need a whole disk ≥ $(( MIN_DISK_BYTES / 1024 / 1024 / 1024 )) GiB; floppy/loop/optical excluded)."

 if [[ "$UNATTENDED" == "1" ]]; then
    TARGET_DISK="${NOMARCHY_DISK:?NOMARCHY_DISK required in unattended mode}"
+    # Unattended still rejects non-installable targets (CI footgun).
+    if ! printf '%s\n' "${disks[@]}" | grep -q "^${TARGET_DISK} "; then
+        fail "$TARGET_DISK is not an installable disk (missing, live medium, or below size floor)."
+    fi
 else
    choice=$(printf '%s\n' "${disks[@]}" \
        | gum choose --header "Install Nomarchy on which disk? (EVERYTHING on it will be erased)")
--- a/pkgs/nomarchy-theme-sync/nomarchy-theme-sync.py
+++ b/pkgs/nomarchy-theme-sync/nomarchy-theme-sync.py
@@ -370,9 +370,23 @@ def run_switch() -> None:
            "Rebuild FAILED — scroll up for last lines; run nomarchy-doctor",
            urgency="critical",
        )
+        # Live ISO pins only the *default* theme's HM generation. Applying
+        # another preset offline tries to build the world from source (#113).
+        offline_hint = ""
+        try:
+            import socket
+            socket.create_connection(("1.1.1.1", 53), timeout=1.5).close()
+        except OSError:
+            offline_hint = (
+                " Offline? Only the live ISO's already-pinned theme is "
+                "guaranteed without a network — other themes need builds "
+                "or a binary cache. Connect and retry, or apply the default "
+                "theme (boreal)."
+            )
        die(
            "rebuild failed (state already written; fix and re-run). "
-            "Diagnose: nomarchy-doctor. Recovery: docs/RECOVERY.md"
+            "Diagnose: nomarchy-doctor. Recovery: docs/RECOVERY.md."
+            + offline_hint
        )
    notify("Changes applied ✓")
    # Waybar runs from Hyprland's exec-once (not a systemd unit HM would
--- a/tools/check-install-safety.py
+++ b/tools/check-install-safety.py
@@ -72,6 +72,25 @@ check(
    "missing online-only compose-lock fallback warn",
 )

+# ── 1b. Disk picker: exclude floppy/pseudo/tiny (#112) ─────────────────
+check(
+    "list_installable_disks" in script,
+    "missing list_installable_disks helper (#112)",
+)
+check(
+    "/dev/fd*" in script or r"/dev/fd*" in script,
+    "disk filter must exclude /dev/fd* (floppy — OVMF lists fd0 first)",
+)
+check(
+    "MIN_DISK_BYTES" in script,
+    "disk filter must enforce a minimum size floor",
+)
+check(
+    "8 * 1024 * 1024 * 1024" in script or "8 * 1024**3" in script
+    or "$((8 * 1024 * 1024 * 1024))" in script,
+    "default min disk size should be 8 GiB (catches fd0 and crumbs)",
+)
+
 # ── 2. Disk signature warn ──────────────────────────────────────────────
 sig_re = re.search(
    r"grep -qiE '([^']+)'\s*<<<\s*\"\$existing_sig\"",
--- a/tools/check-menu-back.py
+++ b/tools/check-menu-back.py
@@ -0,0 +1,94 @@
+#!/usr/bin/env python3
+"""Guard the nomarchy-menu Back/Left contract (BACKLOG #106).
+
+Every *internal* list picker built with `rofi_menu` must emit the shared
+`↩ Back` row (via `back`, `printf … "$BACK"`, or a literal). External
+modi/tools and free-text prompts are listed as exceptions — Esc is their
+back path.
+
+Usage: check-menu-back.py <modules/home/rofi.nix>
+Pure — no rofi runtime. Fails on a list menu that can strand the user.
+"""
+
+from __future__ import annotations
+
+import re
+import sys
+from pathlib import Path
+
+if len(sys.argv) != 2:
+    sys.exit(f"usage: {sys.argv[0]} <rofi.nix>")
+
+text = Path(sys.argv[1]).read_text(encoding="utf-8")
+
+# Strip Nix multi-line comments lightly? Keep simple — work on the whole file.
+# Find `| rofi_menu` call sites (possibly with line breaks before `|`).
+# Look at a window of lines before each call for BACK emission.
+
+lines = text.splitlines()
+
+# Exception markers: if the window mentions these, allow no BACK row.
+EXCEPTION_MARKERS = (
+    "External tool:",
+    "external modi",
+    "Esc closes",
+    "Esc is their back",
+    "free-text",
+    "Free-text",
+    "no list to",
+    "rofi_menu -p search",  # web: empty stdin free-text
+    "rofi_menu -p ask",
+    # calc/emoji/network are exec-only without rofi_menu list construction
+)
+
+def window_has_back(chunk: str) -> bool:
+    # Shared helpers emit ↩ Back: bare `back`, printf of $BACK, or a literal.
+    if re.search(r"(?m)^\s*back\s*$", chunk):
+        return True
+    if re.search(r"\bback\b", chunk) and "back()" not in chunk.split("rofi_menu")[0][-20:]:
+        # `back` as a command (not only the word in comments). Prefer line-start.
+        if re.search(r"(?m)^\s*back\b", chunk):
+            return True
+    if "$BACK" in chunk or r"\$BACK" in chunk or "↩ Back" in chunk:
+        return True
+    if re.search(r'printf[^\n]*BACK', chunk):
+        return True
+    return False
+
+
+def window_is_exception(chunk: str) -> bool:
+    return any(m in chunk for m in EXCEPTION_MARKERS)
+
+
+# Match lines that invoke rofi_menu as a picker (not the function definition).
+invoke_re = re.compile(r"\brofi_menu\b")
+bad: list[str] = []
+for i, line in enumerate(lines):
+    if "rofi_menu()" in line or "rofi_menu ()" in line:
+        continue
+    if not invoke_re.search(line):
+        continue
+    # Skip comments and the definition of the helper's docs.
+    stripped = line.lstrip()
+    if stripped.startswith("#"):
+        continue
+    # Window: up to 40 lines before this call (menu construction).
+    start = max(0, i - 40)
+    chunk = "\n".join(lines[start : i + 1])
+    if window_is_exception(chunk):
+        continue
+    # Free-text: empty stdin piped in
+    if re.search(r"rofi_menu[^\n]*<\s*/dev/null", chunk) or re.search(
+        r"rofi_menu[^\n]*\n[^\n]*<\s*/dev/null", chunk
+    ):
+        continue
+    if not window_has_back(chunk):
+        bad.append(f"line {i + 1}: rofi_menu without ↩ Back in preceding construction:\n  {line.strip()}")
+
+if bad:
+    print("check-menu-back: FAIL — internal list menus must end with ↩ Back:\n", file=sys.stderr)
+    for b in bad:
+        print(b, file=sys.stderr)
+    sys.exit(1)
+
+print(f"check-menu-back: ok — {sum(1 for l in lines if invoke_re.search(l) and 'rofi_menu()' not in l)} rofi_menu sites scanned")