docs(guide): keep the Graphics screenshot after #952's uniform-snapshot declarations #1144
No reviewers
Labels
No labels
area/accessibility
area/audio
area/input
area/install
area/menu
area/renderer
area/saves
area/tooling
area/upscaling
beanstalk: approved
beanstalk: failed
beanstalk: needs approval
beanstalk: running
beanstalk: tested
ci: approved
good first issue
gpu/amd
gpu/igpu
gpu/intel
gpu/nvidia
help wanted
kind
bug
kind
chore
kind
docs
kind
feature
kind
performance
kind
question
needs measurements
platform/linux
platform/macos
platform/steam-deck
platform/windows
preview
priority
critical
priority
high
priority
low
regression
risk
high
risk
low
risk
medium
status
blocked
status
confirmed
status
duplicate
status
invalid
status
needs-info
status
wontfix
ci: approved
No milestone
No project
No assignees
2 participants
Notifications
Due date
No due date set.
Dependencies
No dependencies set
Reference
MagicBeanDev/Longshot!1144
Loading…
Reference in a new issue
No description provided.
Delete branch "TrashPanda/Longshot:docs/guide-retain-test-vram"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Type: docs · Risk: low · Tested on: Linux, game-free checks only · Needs: none
Why
mainis red on the Guide screenshots check. The published Graphics image is still the right first view, butgpu/vk/vk_internal.hmoved when #952 added the uniform-snapshot declarations, and the review record was not updated. Every pull request then has to explain a failure that is not its own.What changed
The Graphics screenshot is kept. Its review record now names the
vk_internal.hthat was compared, as the guide's README asks when a newer source still matches the image.docs/images/guide/manifest.json,settings-graphics.png'sreviewed_sourcetakes the new hash ofgpu/vk/vk_internal.hand asha256:revision of the reviewed file list.review.notessays why it is retained. The PNG, its hash,captured_sourceand the capture fields are unchanged.What was compared: a fresh game-free render of main
a413bf1c, the same menu fixture the image was made with. The first view matches the published PNG: Quality preset selected, Frame generation Off, Frame generation by hidden, Game frame rate cap in view. The window title differs by the revision string only. HUD layout is on the Graphics page below this frame. The only reviewed input whose hash moved isvk_internal.h; the menu fixture does not call the new declarations.Measurements
Verdict: passes: the Guide screenshots check fails on main with one issue and passes with this PR. Nothing to measure in the game.
Hardware: Ubuntu 26.04, Ryzen 7 9800X3D, 30 GB RAM, RTX 4070 Ti (not used: game-free checks)
Scene: none, no game run; the guide's screenshot check
Method:
check_guide_screenshots.py --review-sourcesonce on main and once on this branch (deterministic, so one run each)How to reproduce:
python3 -I -B scripts/check_guide_screenshots.py --review-sources, thentests/test_guide_screenshots.pya413bf1c9a0262f7check_guide_screenshots.py --review-sourcestests/test_guide_screenshots.pyFlags and settings
None.
Testing
Tested: on Linux (Ubuntu 26.04), the Guide screenshots check and
tests/test_guide_screenshots.pyon maina413bf1cand on this commit. The picture was compared by eye with the published PNG.python -I -B tests/run_all.py: 596 passed, 23 failed, 80 skipped, the same 23 that fail on maina413bf1con this PC (test_linux_update,test_make_game_cache,test_crt_options,test_bug_report_defaults,test_damage_debug_menu,test_failed_compile_keeps_objects,test_dlss_feature_size,test_framegen_preset,test_guest_imported,test_hud_layout_row,test_lift_lookups,test_linux_preflight,test_menu_i18n,test_missing_nup,test_mods_size_table,test_pending_guest_import,test_rumble_menu,test_run_all_changed,test_upscaler_rows,test_upscaler_round_trip,test_walk,test_walk_key,test_vk_occlusion_queries). None of them reads this manifest line.Not tested: nothing the game runs.
Must be tested: None.
Risks
What could break: nothing the game runs. If the comparison is wrong, the guide's Graphics picture is out of date.
Turning it off: revert the commit.
Compatibility: Nothing changes.
Checklist
Everyone
type(scope): description) of the status line's type.main.python -I -B tests/run_all.pypasses (any failure that also fails onmainis named).third_party/untouched (a new folder with aPROVENANCE.mdinstead).When it applies
Assisted-by:/Co-Authored-By:trailers.WIP: docs(guide): keep the Graphics screenshot after #1107's vk_internal.h declaration, so main's Guide check passes againto docs(guide): keep the Graphics screenshot after #1107's vk_internal.h declaration, so main's Guide check passes againBasic code-only review of
ad9b1c7f00; informational feedback, with final approval left to another reviewer.The original retention decision looks sound: the header change only declares
test_vram_bytes(), neither the menu nor its fixture calls it, and the recorded hashes match this PR's source. The PNG and capture provenance stay unchanged. The reported 14 passing screenshot tests are appropriate evidence for this narrow metadata change.One integration caveat before merge: current main
31f8d8behas since changed five recorded inputs:menu_i18n.h,vk_api.h,vk_internal.h,vk_menu.cpp, andvk_menu.hundergpu/vk/. This manifest update alone therefore no longer restores the source-review guard on current main. Review those newer UI/fixture changes and record a retention or recapture decision, then rerun the guide check on the integrated branch; do not merely refresh the hashes. Coordinate the same manifest with 1125, 68, and 1103 when integrating them.No tests, builds, or fresh screenshot rendering run in this review.
Confirmed on current main
df6180c3. The Graphics review record onad9b1c7fdoes not match six recorded sources:The review named the first five against main
31f8d8be. vk_texture.h has changed since. This head does not turn the Guide check green on current main. I will look at those six for a retention or a recapture before asking anyone to approve, and a hash refresh is not that review. #1125, #68 and #1103 stay in that pass.The six files are not a retain. Between this branch and current main the Graphics page itself changed:
The recorded Graphics screenshot does not show the new row and still has a row the page no longer has. Updating the review hashes would pass the check on a picture that is wrong. The screenshot has to be taken again from the menu fixture on current main. The other four files (vk_internal.h's readback and snapshot declarations, vk_texture.h's ASTC reuse fields, vk_api.h's option number, the audio playback help) do not draw that page.
This pull request stays open and is not ready for review until that new capture is on it.
The fresh render is in, and it matches the image main already publishes. The first view is the same: Quality preset, Frame generation Off, Frame generation by hidden, Game frame rate cap in view. HUD layout sits further down the Graphics page, outside this frame. The window title is the only pixel difference, and that is the revision string.
The Guide check on main
a413bf1cfails for one file, gpu/vk/vk_internal.h, from #952's uniform-snapshot declarations. The menu fixture does not call them. #1144 now only records that hash. The PNG is unchanged. The guide check passes on9a0262f7. The full suite is 596 passed, 23 failed, 80 skipped, the same 23 as main on this PC.docs(guide): keep the Graphics screenshot after #1107's vk_internal.h declaration, so main's Guide check passes againto docs(guide): keep the Graphics screenshot after #952's uniform-snapshot declarationsSuperseded: main @
544229a2passes check_guide_screenshots.py --review-sources by itself (the Graphics screenshot's review record was updated on main), so this change has nothing left to fix and now conflicts. Reopen if the check goes red for the same reason again.Pull request closed