docs(agent): file #125 ISO still named nixos, #126 display res invert
Some checks failed
Check / eval (push) Has been cancelled

Bernardo: live/install ISO artifact is nixos-live-….iso (image.baseName);
live VM Display picker appeared to invert higher/lower modes — investigate
sort/apply vs QEMU scaling before fixing.
This commit is contained in:
2026-07-15 09:53:14 +01:00
parent 545e40f4d9
commit 25e498a5ef
2 changed files with 68 additions and 0 deletions

View File

@@ -31,6 +31,62 @@ post-boot sessions). Preserve separation: the installer bake failure and the
flake pin are different root causes even when they show up on the same machine. flake pin are different root causes even when they show up on the same machine.
(Terminal / Ghostty-on-Acer → shipped as Kitty-only, #95.) (Terminal / Ghostty-on-Acer → shipped as Kitty-only, #95.)
### 125. Live/install ISO artifact is still named `nixos-…`
**Observed:** a built live image lands as e.g.
`result/iso/nixos-live-26.05.20260710.8f0500b-x86_64-linux.iso` — users (and
USB writers) still see **NixOS**, not Nomarchy.
**Already branded elsewhere:** `isoImage.volumeID = "NOMARCHY_LIVE"`,
`isoImage.edition = "live"`, splash/GRUB theme, `system.nixos.distroName` /
`distroId = "nomarchy"` (os-release). The **filename** comes from
`image.baseName` / `image.fileName` (nixpkgs ISO module; formerly
`isoImage.isoBaseName` / `isoName`), which still default to the NixOS
template prefix.
**Fix direction:** set `image.baseName` (or the current nixpkgs equivalent)
to something like `nomarchy-live` in `hosts/live.nix` (and any install ISO
target if split later). Confirm boot menu strings / isolinux labels if they
still say NixOS after the rename. Guard: a cheap check or docs note that
`result/iso/nomarchy-*.iso` is the expected path (TESTING / README build
snippet).
**Pass:** `nix build …isoImage` produces a `nomarchy-…` ISO filename (not
`nixos-…`); volume label and user-facing boot branding stay Nomarchy.
### 126. Display resolution picker: choosing higher modes makes the image smaller (live VM)
**Observed (Bernardo, live ISO in a VM):** System Display resolution
picker — the **higher** the listed mode he picked, the **lower** the
on-screen resolution *appeared* (inverted / wrong mapping). Live session,
not necessarily install.
**Code path:** `modules/home/rofi.nix` `display)` — builds a mode list from
Hyprland tokens + `hyprctl monitors -j``.availableModes[]`, then
`sort -rV -u`, applies with
`hyprctl keyword monitor "$name,$mode,$pos,$scale"` and persists
`settings.monitors.$name`.
**Likely causes to separate:**
1. **Menu/sort bug** — labels or `sort -rV` on `WxH@R` strings dont match
what Hyprland applies (wrong mode string applied).
2. **QEMU/viewer scaling** — guest mode really increases while the host
window is fixed, so content *looks* smaller (not a Nomarchy bug; document).
3. **Scale factor interaction** — keeping current `.scale` while changing
mode on virtio-gpu yields surprising logical size.
**Fix direction:** reproduce in QEMU with `tools/test-live-iso.sh` (or
manual), log `hyprctl monitors -j` before/after each pick, assert applied
`width`/`height` match the chosen token. Fix sort/format if wrong; if pure
viewer scaling, tell the user and optionally show current mode from
`hyprctl` after apply. Prefer a permanent guard on mode-string formatting
if the bug is in our list/apply path.
**Pass:** picking `1920x1080@…` yields a monitor at 1920×1080 (hyprctl);
picking a lower mode yields lower; menu order is highest-first and honest.
If the only inversion is QEMU window chrome, document it in TESTING.md and
close with that finding — do not “fix” real hardware that was fine.
### 123. Installer HM pre-activate aborts on dconf / missing XDG_RUNTIME_DIR ### 123. Installer HM pre-activate aborts on dconf / missing XDG_RUNTIME_DIR
**Seen on both machines** (Dell XPS 9350 and Acer M5-481T), same log shape at **Seen on both machines** (Dell XPS 9350 and Acer M5-481T), same log shape at

View File

@@ -19,6 +19,18 @@ Template:
--- ---
## 2026-07-15 — File #125 ISO name + #126 display resolution invert (no code)
- **Task:** Bernardo: (1) installer ISO still named nixos; (2) live VM
Display picker — higher modes made the screen look lower-res.
- **Did:** NOW **#125** — artifact is `nixos-live-….iso` because
`image.baseName` is still the nixpkgs default; volumeID already
NOMARCHY_LIVE. **#126** — rofi display mode list/apply path; separate
sort bug vs QEMU scaling before coding.
- **Verified:** V0 (docs/agent only); confirmed `result/iso/nixos-live-…iso`
on tree.
- **Pending:** implement when picked.
- **Next suggestion:** still #123 (HM pre-activate), or #125 if a one-liner.
## 2026-07-15 — #95 complete: Kitty is the only terminal (this commit) ## 2026-07-15 — #95 complete: Kitty is the only terminal (this commit)
- **Task:** Bernardo: make kitty default and only option — drop dual-terminal - **Task:** Bernardo: make kitty default and only option — drop dual-terminal
complexity (Ghostty failed on Acer OpenGL 4.2; fallback was more code). complexity (Ghostty failed on Acer OpenGL 4.2; fallback was more code).