fix: unify filament temp mixing blocking with plate error gating

[iesteem]

Summary

Five commits hardening the filament temperature mixing (high/low temp materials on the same plate) blocking path, plus a related defensive fix.

Commit-by-commit

1. fix: unify filament temp mixing blocking with plate error gating

   • Remove has_sliceable_plate_for_slice_all() hijack in the Preview-tab switch that triggered an unintended "slice all" instead of switching to preview.
   • Add m_filament_temp_blocked flag to PartPlate so can_slice() reflects the filament-temp-mixing state, consistent with m_apply_invalid for object-outside errors.
   • sync_filament_temp_mixing_notification() manages its own flag, independent of the background validation system.
   • Simplify EVT_GLVIEWTOOLBAR_PREVIEW to use can_slice() as a unified gate for both error types.

2. fix: use on-demand checks instead of cached flag for filament temp mixing

   • Replace the per-plate cached m_filament_temp_blocked flag with on-demand is_plate_blocked_by_filament_temp_mixing() checks in the rendering and preview-switch paths.
   • Eliminates staleness risks identified during code review where non-current plates could retain stale blocked state after filament changes, undo/redo, or plate switching during slicing.
   • GLCanvas3D: on-demand red-exclamation check on plate thumbnails; Plater: on-demand check in EVT_GLVIEWTOOLBAR_PREVIEW.

3. refactor: simplify filament slot collection in check_filament_temp_mixing

   • Merge duplicate key lists, rename used_slots for clarity, group logic into scoped blocks. No semantic change.

4. refactor: rename STRING_EXCEPT_FILAMENTS_DIFFERENT_TEMP to STRING_EXCEPT_FILAMENTS_MIXING_TEMP

   • Better reflects that the check is about high/low temperature material mixing, not just "different" temperatures.
   • Touches PrintBase.hpp, Snapmaker_Orca.cpp (CLI), and Plater.cpp.

5. feat: show per-filament high/low temp breakdown in mixing notifications

   • Plater::check_filament_temp_mixing now exposes a FilamentTempMixingDetail (high/low 1-based slot lists) via a new overload.
   • Notification text becomes dynamic; sync_filament_temp_mixing_notification caches the exact pushed strings to work around NotificationManager's exact-text close matching.
   • Added slice-all variants listing every conflicting plate.

Related defensive fix (bundled)

• GCodeReader.cpp: parse_file_raw_internal() returns false when fopen fails, preventing fread(NULL) from triggering a CRT invalid-parameter crash.
• Localization: zh_CN/Snapmaker_Orca_zh_CN.po updated with the new dynamic notification strings (Chinese only; other languages to be synced by CI / update-translation.yml).

Downstream effects

Red exclamation icon on plate thumbnails, disabled slice/export buttons, and blocked auto-slice on preview switch now work uniformly for both object-outside and filament-temp-mixing errors.

---

Reviewer Focus

Please prioritize the following (ordered by risk):

1. On-demand vs cached flag tradeoff (most important)

Commit 2 reverted the cached flag from commit 1 in favor of on-demand checks. Please confirm:

• is_plate_blocked_by_filament_temp_mixing() is cheap enough to call on every plate-thumbnail render (GLCanvas3D::_render_imgui_select_plate_toolbar) — it walks model().objects and resolves presets each call. For projects with many plates / objects, is there measurable render-time impact?
• All read sites that previously trusted m_filament_temp_blocked now go through the on-demand check. Are there any remaining readers of the (now-removed) flag that would silently compile to false?
• Was the cached flag kept anywhere (e.g. for PartPlate::can_slice())? If can_slice() no longer reflects temp-mixing state, the "unified gating" promise of commit 1 may regress.

2. EVT_GLVIEWTOOLBAR_PREVIEW behavior change

Before: clicking Preview with a sliceable plate auto-fired slice_all.
After: blocking checks run first; if blocked, switch to Preview view without slicing.
Please confirm the non-blocking path still auto-slices as before, and that the workflow change is intentional and surfaced to users (release notes / changelog).

3. Notification close-then-push correctness

sync_filament_temp_mixing_notification() now caches filament_temp_mixing_last_error_text / filament_temp_mixing_last_warning_text for exact-text close. Please check:

• All early-return paths (e.g. curr_plate == nullptr) — do they leak a stale notification by skipping the close step?
• Across the three triggers (plate switch, filament change, bed-type change), the previous notification is always cleared before pushing the new one.
• Slice-all mode caches a different text shape than single-plate mode; verify the close path matches the right cache when switching modes.

4. Slot-collection refactor (commit 3) is purely cosmetic

"No semantic change" is claimed. Please diff-check:

• The merged key list covers exactly what the two original lists covered (wall_filament, sparse_infill_filament, solid_infill_filament, support_filament, support_interface_filament, wipe_tower_filament).
• The uses_default_extruder branch + global extruder resolution order is unchanged.
• The 1-based → 0-based slot conversion is preserved.

5. Enum rename impact

STRING_EXCEPT_FILAMENTS_DIFFERENT_TEMP → STRING_EXCEPT_FILAMENTS_MIXING_TEMP. Search the whole repo (including 3MF serialization, Sentry reporting, CLI) to confirm:

• No missed call sites.
• No persistence layer stores this enum by name (only by integer value = 2, which is preserved).

6. Per-filament breakdown formatting

The notification text is now dynamic and lists slots like [1] PLA White; [3] PETG Black. Please confirm:

• Slot labels match what the user actually sees (preset name resolution, 1-based indexing).
• Translation still works for the surrounding template — i.e. dynamic slot lists don't break gettext format strings (%1% etc.).
• For very long slot lists (10+ filaments), does the notification render sanely (no truncation / layout break)?

7. Bundled GCodeReader.cpp fix

• The fopen NULL check is a defensive fix unrelated to filament temp mixing — confirm it's intentional to bundle here (vs. a separate PR).
• Does any Sentry crash fingerprint depend on this path crashing? If so, the fix will silence those reports.

8. i18n coverage

Only zh_CN.po is updated; the other 20 languages will fall back to English for the new dynamic strings. Is that acceptable for merge, or should we wait for update-translation.yml to sync?

---

Test Plan

• Plate with both high-temp and low-temp filaments in use → ValidateError fires, slice button greyed out, red exclamation on plate thumbnail
• Click Preview tab on a blocked plate → switches to Preview without auto-slicing; notification remains visible
• Click Preview tab on a compatible plate → still auto-slices via slice_all (no regression)
• Resolve the mixing conflict (swap a filament) → notification clears immediately, slice button re-enables, exclamation disappears (no stale state)
• Undo/redo a filament swap → blocking state tracks correctly (verifies the on-demand check, not a stale cached flag)
• Switch plates while slicing → non-current plates don't retain stale blocked state
• Slice All with multiple conflicting plates → notification lists each plate with its high/low slot breakdown
• Slice All with one blocked / one clean plate → blocked plate skipped, clean plate slices
• Notification text in zh_CN renders correctly; English / other locales render with English fallback (no crash, no format-string breakage)
• CLI slice on a mixing-conflict project → still reports the error via CLI_FILAMENTS_DIFFERENT_TEMP (rename is internal-only)
• Loading a G-code file whose path is invalid → returns cleanly, no CRT crash

---

🤖 Generated with Claude Code

[iesteem]

Review findings addressed

Pushed two commits to fix_top_cover addressing the blocker/major review findings:

• 1fc3eb9 — B1/B2/B3/B4 + m1 fixes:
  • B1 (perf): added Plater::priv::filament_temp_mixing_blocked_cache + is_plate_blocked_by_filament_temp_mixing_cached(). Cache is refreshed once per sync_filament_temp_mixing_notification() call (which already runs at every state-changing event), so the per-frame plate-thumbnail render path is now O(1).
  • B2 (notification leak): sync_filament_temp_mixing_notification() now closes previously-pushed notifications BEFORE the curr_plate == nullptr early return.
  • B3 (tr_u8): removed. I18N::translate_utf8() returns std::string by value — the dangling-pointer narrative in the helper's comment was incorrect. All call sites revert to _u8L.
  • B4 (warning close mismatch): the warning notification body was being pushed with a WARNING: prefix but close_validate_warning_notification(text) re-prepends the same prefix, so the close never matched. Fixed by caching the un-prefixed body.
  • m1 (style): restored Allman brace style in new helpers and the check_filament_temp_mixing overload to match the rest of Plater.cpp.
• abe6eb9 — M3 scope split:
  • Reverted GCodeReader.cpp (NULL FILE* check) and Plater::load_gcode (wxFileExists early-reject) to main state, removing them from this PR's diff.
  • These two defensive fixes now live in their own PR: fix: prevent NULL FILE* crash when loading a missing G-code file #648. They have an independent regression surface (loading missing/invalid G-code) and should ship separately for clean bisect.

Remaining items for the author / reviewers

• Feature: Changename to "Snapmaker_Orca" #2 EVT_GLVIEWTOOLBAR_PREVIEW workflow change — please confirm the non-blocking auto-slice path still behaves as before (the new can_slice() gate may change behavior on previously auto-slicing plates).
• feature: change version to 0.0.0, change release_id back to nightly build #5 enum rename — confirmed no remaining STRING_EXCEPT_FILAMENTS_DIFFERENT_TEMP references in the tree; no 3MF persistence by name. ✓
• feature:fix engine #8 i18n coverage — only zh_CN is updated for the new dynamic strings. Acceptable for merge if other locales fall back to English cleanly; otherwise wait for update-translation.yml sync.

Test plan additions

Please add manual verification for:

• AllowedWarning state → switch plates → warning banner disappears instantly (verifies B4 fix).
• Multi-plate project (10+ plates, many objects) → drag 3D view → no frame drops (verifies B1 fix).
• curr_plate transiently null during plate switch → no stale error banner remains (verifies B2 fix).