fix(macos): keychain prompts, session persistence, texture apply, mesh-tex generation - #718
Merged
Merged
Conversation
Two startup/usability fixes: 1. macOS keychain prompt spam. CloudAccountMenuButton::refresh() reads the cloud session 3-4 times during startup (migrate + hasSession + cloudDisplayName), and on macOS every SecItemCopyMatching can raise a "QtMeshEditor wants to use confidential information" dialog. Add an in-process cache to CloudCredentialStore so the OS keychain/credential store is queried at most once per process. Save primes the cache, clear sets it to the empty state, so post-sign-in/out refreshes never re-hit the OS. Also set kSecAttrAccessibleAfterFirstUnlock so the item is readable without an interactive unlock each session. 2. "Applying texture to the model does nothing" for materials whose diffuse TUS has a numeric/empty name (Assimp-imported "0" slot, user-added unit). RTShaderHelper::wirePbrSlotsForFFP only wired NAMED slots, so the diffuse never got its FFP modulate op and the texture loaded but never rendered. Add a fallback: when a pass exposes no recognized albedo/diffuse_map slot, the first plain textured slot receives the diffuse modulate op. Fixes the manual setTextureName path, plain txt2img, and the MCP set_texture path in one place. Tests: cache-serves-reads and corrupt-file cache-reset coverage. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
…h-tex generation
Several related macOS / Material-Editor fixes reported in testing:
1. Repeated macOS keychain prompts at startup. CloudCredentialStore no longer
uses the OS secret store (Keychain / Credential Manager / libsecret) — every
macOS keychain read raised a confirmation dialog and the account control reads
the session several times at launch. The session now lives in QSettings (the
per-user preference store, no prompt). Removes the Security.framework / wincred
/ libsecret code, the CloudCredentialStore_linux.* files, and their CMake/link
wiring. Keeps an in-process cache to collapse the startup reads.
2. Cloud session not persisting (user logged out each launch). The device-code
sign-in handler removed the cloudToken/expiresAt/email QSettings keys right
after saveSession() wrote them — leftover keychain-era cleanup. Those keys are
now the storage, so the removes are dropped.
3. Applying a texture didn't show on the model (only the preview updated):
- RTShaderHelper::wirePbrSlotsForFFP now wires an unnamed/numeric diffuse TUS
(e.g. an Assimp-imported "0" slot) as the diffuse fallback when no
albedo/diffuse_map slot exists — previously only named slots got the FFP
colour-op, so the texture loaded but never rendered.
- MaterialEditorQML::ensureTextureInMaterialGroup loads the chosen texture
into the MATERIAL's resource group before binding, so the RTSS-rendered
mesh can resolve it (it was resolving against the wrong group → yellow
placeholder while the disk-reading preview looked fine).
- updateMaterialText re-applies via SubEntity::setMaterial(ptr) instead of
setMaterialName(sameName), which could keep a stale technique pointer.
4. Plain AI (txt2img) texture not applying: onSDGenerationCompleted now routes
the generated texture through setTextureName() (group resolution + RTSS
rebind) instead of a raw texUnit->setTextureName().
5. "Use selected mesh (depth-conditioned)" never started: the QML called
MaterialEditorQML.hasSelectedMesh() but it is exposed as a Q_PROPERTY, so the
call threw "not a function" and silently aborted the handler. Read it as a
property. Also: missing-ControlNet now emits a non-fatal sdGenerationNotice
(was sdGenerationError, which reset the in-flight run) so generation proceeds
in prompt-only mode. The "Use selected mesh" checkbox no longer unchecks
itself when generation starts (auto-uncheck now keys off selection loss, not
the disabled state).
6. Logged-out account icon blank in deployed builds: the avatar used an SVG
QIcon that renders empty without the Qt SVG icon-engine plugin (showed
locally, vanished in the installed app). The avatar glyph is now drawn with
QPainter for both states — no plugin dependency.
7. Docs: Homebrew install now includes `brew trust fernandotonon/qtmesheditor`
(Homebrew now requires trusting third-party taps before loading their casks).
Tests updated for the QSettings-backed CloudCredentialStore.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
Warning Review limit reached
More reviews will be available in 48 minutes and 29 seconds. Learn how PR review limits work. Your organization has run out of usage credits. Purchase more credits in the billing tab to continue. ⌛ How to resolve this issue?After more reviews become available, a review can be triggered using the We recommend that you space out your commits to avoid hitting the rate limit. 🚦 How do rate limits work?CodeRabbit enforces hourly rate limits for each developer per organization. Our paid plans include higher PR review limits than trial, open-source, and free plans. In all cases, reviews become available again over time. During sustained high-volume PR review activity, CodeRabbit may temporarily slow when the next review becomes available. Please see our Fair Usage Limits Policy for further information. 📝 WalkthroughWalkthroughThis PR migrates QtMeshEditor's cloud session storage to per-user QSettings with an in-process cache, drops libsecret wiring, updates tests and CMake, refactors avatar rendering, improves material/RTSS texture resolution and Stable Diffusion flows, refines PBR diffuse fallback logic, and updates macOS packaging/docs and pinned release refs to 3.5.3. ChangesCloud Credential Storage Migration & Rendering Improvements
Estimated code review effort🎯 4 (Complex) | ⏱️ ~60 minutes Possibly related PRs
Poem
🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (4 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 47aa978d59
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
Comment on lines
+103
to
+106
| // Tokens already live in QSettings under the same keys, so there is nothing | ||
| // to migrate. Retained as a no-op so existing callers (the account control's | ||
| // refresh) keep compiling, and as the hook point should the storage backend | ||
| // change again. |
There was a problem hiding this comment.
For users upgrading from a build that saved the cloud session in Keychain/Credential Manager/libsecret, this no-op leaves the only stored token behind in the old backend while loadSession() now reads only QSettings. In that upgrade path the account control calls this migration hook and then hasSession(), so previously signed-in users are treated as logged out until they authenticate again; the migration should read the old secret/fallback location once and copy it into the new QSettings keys before dropping the platform-store code.
Useful? React with 👍 / 👎.
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 4
🧹 Nitpick comments (1)
src/CMakeLists.txt (1)
505-507: ⚡ Quick winRemove the stale Debian
libsecretdependency too.This block says the secret-store libraries are gone, but
DEBIAN-control.instill declareslibsecret-1-0. The Linux package will keep pulling the removed dependency unless that metadata is updated alongside this cleanup.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@src/CMakeLists.txt` around lines 505 - 507, The comment in CMakeLists.txt indicates platform secret stores were removed but packaging metadata still lists libsecret; update the Debian package metadata by removing the libsecret-1-0 dependency from DEBIAN-control.in (and any other packaging control files referencing libsecret-1-0) so the produced .deb no longer pulls the stale library; ensure you also update any packaging-related variables or comments that mention secret-store linking to avoid future confusion.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@src/CloudCredentialStore.cpp`:
- Around line 40-50: writeToSettings currently calls QSettings::sync() but
ignores failures; update writeToSettings to check QSettings::status() after
settings.sync() and return a bool indicating success (true when status ==
QSettings::NoError). Then change saveSession (the caller that currently returns
true if session.token is non-empty) to call writeToSettings and propagate its
boolean result (return false if writeToSettings reports an error), ensuring both
places mentioned (the token branch and the other branch around lines 63–70) use
the new return value so MainWindow::signInToQtMeshCloud() receives failure when
sync fails.
- Around line 30-37: The readFromSettings() function populates email and
expiresAt even when there is no token, returning a partial CloudSession; change
readFromSettings() to treat a missing/empty
settings.value(AppSettingsKeys::cloudToken()) as an empty session by checking
the token first and returning the default CloudSession immediately if token is
empty, only reading AppSettingsKeys::cloudTokenExpiresAt() and
AppSettingsKeys::cloudUserEmail() when a non-empty token exists so loadSession()
cannot return stale identity data.
In `@src/MaterialEditorQML.cpp`:
- Around line 1008-1076: In ensureTextureInMaterialGroup, the new disk-read path
that loads the image into Ogre should record a Sentry breadcrumb for
file.import; after successfully loading the bytes and calling
tm.loadImage(texStd, matGroup, image) add
SentryReporter::addBreadcrumb("file.import", {{"path", srcPath}, {"status",
"ok"}, {"name", texStd}}) and in the catch/all-failure branch (or just before
each early return when srcPath is missing or file open fails) add a failure
breadcrumb like SentryReporter::addBreadcrumb("file.import", {{"path", srcPath},
{"status", "failed"}, {"reason","not found"/"io error"}}) so texture-resolution
diagnostics consistently log imports from ensureTextureInMaterialGroup.
In `@src/RTShaderHelper.cpp`:
- Around line 503-505: The current diffuse detection uses literal checks
("albedo" || "diffuse_map") which duplicates logic and misses aliases defined
elsewhere; replace those manual string checks with the existing helper
isAlbedoSlotName(n) so hasNamedDiffuse is set when any albedo alias (e.g.
"Diffuse", "BaseColor", "albedo", "diffuse_map") is present; update both
occurrences around the hasNamedDiffuse assignment (the block that sets
hasNamedDiffuse = true and the similar check later) to call isAlbedoSlotName(n)
instead of comparing to hardcoded strings.
---
Nitpick comments:
In `@src/CMakeLists.txt`:
- Around line 505-507: The comment in CMakeLists.txt indicates platform secret
stores were removed but packaging metadata still lists libsecret; update the
Debian package metadata by removing the libsecret-1-0 dependency from
DEBIAN-control.in (and any other packaging control files referencing
libsecret-1-0) so the produced .deb no longer pulls the stale library; ensure
you also update any packaging-related variables or comments that mention
secret-store linking to avoid future confusion.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro
Run ID: 2b9b88f2-1917-4bcf-80f8-5c101db7a03f
📒 Files selected for processing (15)
README.mdqml/TexturePropertiesPanel.qmlsrc/CMakeLists.txtsrc/CloudAccountMenuButton.cppsrc/CloudCredentialStore.cppsrc/CloudCredentialStore.hsrc/CloudCredentialStore_linux.csrc/CloudCredentialStore_linux.hsrc/CloudCredentialStore_test.cppsrc/MaterialEditorQML.cppsrc/MaterialEditorQML.hsrc/RTShaderHelper.cppsrc/mainwindow.cppwebsite/src/DocsApp.jsxwebsite/src/data/content.js
💤 Files with no reviewable changes (2)
- src/CloudCredentialStore_linux.h
- src/CloudCredentialStore_linux.c
The update-homebrew-cask release step appended a `postflight do...end` block to the END of the cask file with `>> "$CASK_FILE"`. The cask already had a postflight inside its `cask do...end`, so the appended block landed AFTER the closing `end` — Homebrew parsed it as a top-level `postflight` call and rejected the cask with "undefined method 'postflight'", breaking every install. - Remove the broken append; the canonical cask already does lsregister inside its single postflight. - Add packaging/macos/qtmesheditor.rb.in as the canonical template (modern double-quote DSL, one postflight: quarantine clear + lsregister, both CLI binaries, correct com.none.QtMeshEditor zap path). CI rewrites the whole cask from it if an older tap revision is missing lsregister, instead of appending. The tap repo (fernandotonon/homebrew-qtmesheditor) was fixed directly so the current 3.5.2 release installs again. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
…in CI - Bump project version 3.5.2 -> 3.5.3 (CMakeLists.txt, single source of truth); synced pinned doc refs via scripts/sync-doc-versions-from-cmake.sh. - update-homebrew-cask CI job now regenerates the ENTIRE cask from packaging/macos/qtmesheditor.rb.in on every release instead of sed-patching / appending: checks out this repo for the template, substitutes @Version@ / @sha256@, and overwrites the tap's Casks/qtmesheditor.rb. This removes both the stray-second-postflight bug (#718) and the stale single-quote sed patterns that would no longer match the double-quote cask DSL. Verified the generated cask passes `ruby -c`. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- CloudCredentialStore: migrate pre-QSettings sessions. migrateLegacySettings- IfNeeded now does a one-time read of the old OS secret store (macOS Keychain / Windows Credential Manager) + the legacy mode-0600 fallback file and copies it into QSettings, so users who signed in on an older build aren't logged out. Re-added the macOS Security / Windows advapi32 link for that read; libsecret stays dropped (no Linux read-back). (Codex P2) - readFromSettings: return an empty session when no token is stored instead of hydrating email/expiry from stray keys. (CodeRabbit) - saveSession: propagate QSettings::sync() failure (writeToSettings now returns status == NoError) so a non-persistent login is reported, not silently swallowed. (CodeRabbit major) - RTShaderHelper::wirePbrSlotsForFFP: use the existing isAlbedoSlotName() helper (covers "Diffuse"/"BaseColor" too) instead of literal "albedo"/"diffuse_map" checks, so the unnamed-diffuse fallback can't mis-wire a pass that uses an alias. (CodeRabbit major) - ensureTextureInMaterialGroup: add file.import Sentry breadcrumbs on success/failure. (CodeRabbit) - DEBIAN-control.in: drop the now-unused libsecret-1-0 dependency. (CodeRabbit) Tests: added legacy-fallback-file migration + tokenless-settings coverage. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Owner
Author
|
Addressed all review feedback in 175fbfa:
Added tests for the legacy-fallback-file migration and the tokenless-settings case. All changed files compile; the only local link failure is the pre-existing |
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Nitpick comments (2)
.github/workflows/deploy.yml (1)
2078-2081: 💤 Low valueOptional: Add
persist-credentials: falsefor read-only template checkout.This checkout is read-only (used only to access the canonical cask template), so following security best practices, add
persist-credentials: falseto prevent the workflow token from persisting in the checked-out repository.🛡️ Suggested enhancement
- name: Check out QtMeshEditor (for the canonical cask template) uses: actions/checkout@v4 with: path: qtmesheditor-src + persist-credentials: false🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In @.github/workflows/deploy.yml around lines 2078 - 2081, The checkout step named "Check out QtMeshEditor (for the canonical cask template)" using actions/checkout@v4 should be made read-only: add the persist-credentials: false key to that step so the workflow token is not persisted into the checked-out qtmesheditor-src repository; update the checkout step configuration to include persist-credentials: false alongside the existing path: qtmesheditor-src.src/CloudCredentialStore.cpp (1)
150-163: ⚡ Quick winAdd Sentry breadcrumbs for session persist/migration paths.
saveSession()andmigrateLegacySettingsIfNeeded()handle user-visible auth persistence and migration outcomes, but these significant operations are not breadcrumbed. Please addSentryReporter::addBreadcrumb()on success/failure paths (file.exportfor settings writes,file.importfor legacy migration reads/writes).Suggested patch sketch
bool CloudCredentialStore::saveSession(const CloudSession& session) { if (!session.hasToken()) return false; if (!writeToSettings(session)) { + SentryReporter::addBreadcrumb(QStringLiteral("Cloud session persist failed"), + QStringLiteral("file.export")); // The write didn't persist (permissions / disk). Don't claim success or // prime the cache, so the caller can surface the failure. invalidateCache(); return false; } + SentryReporter::addBreadcrumb(QStringLiteral("Cloud session persisted"), + QStringLiteral("file.export")); g_sessionCache = session; g_cacheValid = true; return true; } @@ void CloudCredentialStore::migrateLegacySettingsIfNeeded() { @@ const CloudSession legacy = readLegacySecretStore(); - if (legacy.hasToken() && writeToSettings(legacy)) { + if (legacy.hasToken() && writeToSettings(legacy)) { + SentryReporter::addBreadcrumb(QStringLiteral("Legacy cloud session migrated"), + QStringLiteral("file.import")); g_sessionCache = legacy; g_cacheValid = true; + } else if (legacy.hasToken()) { + SentryReporter::addBreadcrumb(QStringLiteral("Legacy cloud session migration failed"), + QStringLiteral("file.import")); } }As per coding guidelines, add breadcrumbs for all user-facing/significant operations via
SentryReporter::addBreadcrumb()with the prescribed categories.Also applies to: 193-207
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@src/CloudCredentialStore.cpp` around lines 150 - 163, Add Sentry breadcrumbs to record success/failure of session persistence and legacy migration: inside saveSession (around writeToSettings success/failure paths) call SentryReporter::addBreadcrumb() with category "file.export" and include context (e.g., outcome "write_success" or "write_failure", and any error/permission info) on both the branch where writeToSettings returns false (before invalidateCache) and where it succeeds (after g_sessionCache/g_cacheValid are set); likewise, in migrateLegacySettingsIfNeeded add SentryReporter::addBreadcrumb() calls with category "file.import" for the legacy-read attempt and for the migration write outcome (both success and failure), including minimal context fields to indicate the step and result so these events are breadcrumbed for later diagnostics.Source: Coding guidelines
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@src/CMakeLists.txt`:
- Around line 527-532: The UnitTests target needs the same platform-specific
legacy secret-store linkage as the main app: update CMakeLists to also call
target_link_libraries for the UnitTests target (in addition to
${CMAKE_PROJECT_NAME}) on Apple and WIN32 so UnitTests links "-framework
Security" on macOS and advapi32 on Windows; this ensures
CloudCredentialStore.cpp (built via TEST_SOURCES) and
CloudCredentialStore_test.cpp which exercises
CloudCredentialStore::migrateLegacySettingsIfNeeded() will link correctly.
---
Nitpick comments:
In @.github/workflows/deploy.yml:
- Around line 2078-2081: The checkout step named "Check out QtMeshEditor (for
the canonical cask template)" using actions/checkout@v4 should be made
read-only: add the persist-credentials: false key to that step so the workflow
token is not persisted into the checked-out qtmesheditor-src repository; update
the checkout step configuration to include persist-credentials: false alongside
the existing path: qtmesheditor-src.
In `@src/CloudCredentialStore.cpp`:
- Around line 150-163: Add Sentry breadcrumbs to record success/failure of
session persistence and legacy migration: inside saveSession (around
writeToSettings success/failure paths) call SentryReporter::addBreadcrumb() with
category "file.export" and include context (e.g., outcome "write_success" or
"write_failure", and any error/permission info) on both the branch where
writeToSettings returns false (before invalidateCache) and where it succeeds
(after g_sessionCache/g_cacheValid are set); likewise, in
migrateLegacySettingsIfNeeded add SentryReporter::addBreadcrumb() calls with
category "file.import" for the legacy-read attempt and for the migration write
outcome (both success and failure), including minimal context fields to indicate
the step and result so these events are breadcrumbed for later diagnostics.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro
Run ID: ba953e4e-f7d2-4d81-83cc-9a620dc33a25
📒 Files selected for processing (11)
.github/workflows/deploy.ymlCMakeLists.txtDEBIAN-control.inREADME.mdpackaging/macos/qtmesheditor.rb.insrc/CMakeLists.txtsrc/CloudCredentialStore.cppsrc/CloudCredentialStore_test.cppsrc/MaterialEditorQML.cppsrc/RTShaderHelper.cppwebsite/src/hooks/useQtmeshActionRef.js
✅ Files skipped from review due to trivial changes (3)
- website/src/hooks/useQtmeshActionRef.js
- CMakeLists.txt
- README.md
🚧 Files skipped from review as they are similar to previous changes (3)
- src/RTShaderHelper.cpp
- src/CloudCredentialStore_test.cpp
- src/MaterialEditorQML.cpp
- Mirror the legacy secret-store linkage on the UnitTests target: it compiles CloudCredentialStore.cpp (which CloudCredentialStore_test.cpp exercises via migrateLegacySettingsIfNeeded -> readLegacySecretStore), so macOS/Windows test builds need "-framework Security" / advapi32 or they fail to link. (CodeRabbit) - Gate PS1-runtime *_test.cpp the same way as their impl: those sources are only added to SRC_FILES when ENABLE_PS1_RIP is ON, but the _test.cpp files were globbed unconditionally — so a tests build with ENABLE_PS1_RIP=OFF linked e.g. TextureDecoder_test.cpp against a missing qHash(TextureDecoder::MaterialKey) and failed. Skip src/PS1/runtime/*_test.cpp when ENABLE_PS1_RIP is OFF. UnitTests now links cleanly locally (macOS, PS1 rip off). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Owner
Author
|
Addressed in 3705dc4: added the |
|
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Fixes a cluster of macOS / Material-Editor issues found while testing the AI texture features.
What was broken & fixed
Repeated macOS keychain prompts at startup (4×).
CloudCredentialStorestored the cloud session in the OS secret store (Keychain/Credential Manager/libsecret); every macOS keychain read raises a prompt and the account control reads the session several times at launch. → Session now stored in QSettings (per-user prefs, no prompt). Removed the Security.framework/wincred/libsecret code, theCloudCredentialStore_linux.*files, and their CMake/link wiring. Kept an in-process cache.Cloud session not persisting (logged out each launch). The sign-in handler removed the
cloudToken/expiresAt/emailQSettings keys right aftersaveSession()wrote them (leftover keychain-era cleanup). → Removes dropped.Applying a texture didn't show on the model (only the preview updated):
RTShaderHelper::wirePbrSlotsForFFPnow wires an unnamed/numeric diffuse TUS (e.g. an Assimp-imported"0"slot) as the diffuse fallback when there's no namedalbedo/diffuse_mapslot — previously only named slots got the FFP colour-op, so the texture loaded but never rendered.MaterialEditorQML::ensureTextureInMaterialGrouploads the chosen texture into the material's resource group before binding, so the RTSS-rendered mesh can resolve it (was resolving against the wrong group → yellow placeholder while the disk-reading preview looked fine).updateMaterialTextre-applies viaSubEntity::setMaterial(ptr)instead ofsetMaterialName(sameName)(stale technique pointer).Plain AI (txt2img) texture not applying:
onSDGenerationCompletednow routes the result throughsetTextureName()(group resolution + RTSS rebind) instead of a rawtexUnit->setTextureName()."Use selected mesh (depth-conditioned)" never started: QML called
MaterialEditorQML.hasSelectedMesh(), but it's aQ_PROPERTY→ the call threw "not a function" and silently aborted the handler. Now read as a property. Missing-ControlNet emits a non-fatalsdGenerationNotice(wassdGenerationError, which reset the run) so it proceeds in prompt-only mode. The checkbox no longer unchecks itself when generation starts (auto-uncheck keys off selection loss, not the disabled state).Logged-out account icon blank in deployed builds: the avatar used an SVG
QIconthat renders empty without the Qt SVG icon-engine plugin (showed locally, vanished in the installed app). The glyph is now drawn withQPainterfor both states — no plugin dependency.Docs: Homebrew install now includes
brew trust fernandotonon/qtmesheditor(Homebrew now requires trusting third-party taps before loading their casks — this is the new expected workflow, not a cask bug).Testing
CloudCredentialStore_test.cppupdated for the QSettings backend (round-trip, cache, no-op migration).🤖 Generated with Claude Code
Summary by CodeRabbit
New Features
Bug Fixes
Documentation
Chores
Tests