Skip to content

Launcher suite: test_re9's uninstall check follows the file-ownership rule (#219) - #222

Merged
phantoslayer21 merged 3 commits into
phantoslayer21:mainfrom
j0nsh1n:test/suite-ownership-checks
Nov 15, 2026
Merged

phantoslayer21 merged 3 commits into
phantoslayer21:mainfrom
j0nsh1n:test/suite-ownership-checks

Conversation

@j0nsh1n

@j0nsh1n j0nsh1n commented Nov 15, 2026 •

Copy link
Copy Markdown
Contributor

What this changes and why

Part of #219. Test change only; no launcher code changes.

  • test_re9: the test drops the mirror's _storage_\dxgi.dll from the record, so the repair takes that unrecorded copy for the player's and backs it up, and the uninstall puts it back (the file-ownership rule). The check now expects it.
  • run_launcher_tests.py is at 262132 bytes on main, under the 256 KB source limit by 12 bytes, so test_edition_download moves unchanged to launcher_edition_cases.py, the way launcher_ignore_cases.py did. A two-line test_edition_download keeps --only edition_download working.

Dropped after review: the test_rr check and rr.h's header are back to main's text. They change with the WrongGPU\state\ design in 0.6.5.

What each new or substantially changed file is based on

launcher_edition_cases.py: main's test_edition_download, moved byte for byte. The check: own design, following the README rule.

How it was tested

  • With main's own launcher (MinGW WRONGGPU_TESTING build, -DWRONGGPU_PRESET_K=OFF) under Wine (GE-Proton 11-11), before the merge with main: --only re9 goes from 5 failures to 4, the uninstall-after-update check passing. The other 4 are the Wine run's own (empty UI dumps, rr-prepare without a helper), the same on main.
  • After the merge with main (Launcher: the MinGW build of WrongGPU and the suite's --only (#219) #227): the file parses, the moved section's body is byte-identical to main's, check_source_size.py reports 256799 bytes, reuse lint passes. The Wine run was not repeated.
  • Not run: on Windows.

Checklist

  • I make the clean-room statement of CONTRIBUTING.md for this and all my contributions (required once, in my first pull request). (Made in Launcher: install for Proton games from under Wine #105.)
  • Every commit has a Signed-off-by: line (git commit -s; Developer Certificate of Origin 1.1).
  • New source files carry the SPDX licence line for MIT, or are covered in REUSE.toml.

🤖 Generated with Claude Code

… the file-ownership rule

Two checks of tests/run_launcher_tests.py expected what the launcher did
before 0.6.1's file-ownership rule (README, record.h): a file at a recorded
path whose bytes are not the ones WrongGPU recorded stays where it is.

- test_rr: after the in-game switch edits WrongGPU\wronggpu.ini, the next
  repair backs that file up as changed since the install; the plugin
  rewrites it again, so the uninstall leaves it and keeps the backed-up
  generation in WrongGPU-originals. The check now expects that copy (one
  wronggpu.ini there and nothing else, holding the in-game switch's rr_mode).
- test_re9: the test drops the mirror's _storage_\dxgi.dll from the record;
  the repair takes that unrecorded copy for the player's and backs it up, so
  the uninstall puts it back. The check now expects it.

rr.h's header still said the RR settings files go at uninstall after the
plugin rewrote them; it now says what the rule does.

The file was 69 bytes below the 256 KB source limit, so test_edition_download
moves unchanged to launcher_edition_cases.py (as launcher_ignore_cases.py
did); a two-line test_edition_download keeps --only edition_download.

Verified with main's own launcher (MinGW WRONGGPU_TESTING build,
-DWRONGGPU_PRESET_K=OFF) under Wine (GE-Proton 11-11): --only rr goes from
10 to 9 failures and --only re9 from 5 to 4, the two checks now passing;
the rest are the Wine run's own (empty UI dumps, rr-prepare without a
helper). The moved section's body is byte-identical to main's; under the
Wine adapter it stops at the same zipfile path step on main and here.
reuse lint and check_source_size pass (run_launcher_tests.py 257355 bytes).

Part of phantoslayer21#219 (its MinGW and --only re9 parts are a separate PR).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Jonathan Shin <85975340+j0nsh1n@users.noreply.github.com>
@phantoslayer21

Copy link
Copy Markdown
Owner

The test_re9 change is right: the test removes the mirror's dxgi.dll from the record itself, so treating that copy as the player's is the ownership rule doing its job.

test_rr is different. wronggpu.ini is WrongGPU's own settings file, and the edit comes from WrongGPU's own plugin, so uninstall should remove it and leave no WrongGPU-originals folder behind. A file the player edits by hand still stays. That's a launcher change: the plugin's writes need to keep the file recorded as WrongGPU's. It's being done on our side for 0.6.5 with tests for both cases. Could you drop the test_rr part from this PR (or leave it as is until that lands, and we'll rebase)? The re9 part and the rr.h header fix can go in on their own.

@phantoslayer21

Copy link
Copy Markdown
Owner

Update on test_rr: the fix for #219 is changing shape. Neither the plugin nor the launcher will rewrite WrongGPU\wronggpu.ini after install any more. Settings changed in the launcher or the in-game menu go to a WrongGPU-owned WrongGPU\state\ folder that overrides wronggpu.ini and is always removed at uninstall. wronggpu.ini stays install defaults plus whatever the player edits by hand. So test_rr will expect a clean uninstall with nothing kept, and your test_re9 change is unaffected.

j0nsh1n and others added 2 commits November 14, 2026 19:04
Signed-off-by: Jonathan Shin <85975340+j0nsh1n@users.noreply.github.com>
Per the review of phantoslayer21#222: the test_rr check and rr.h's header go back to
main's text. WrongGPU's own settings file is to be removed at uninstall
once settings move to WrongGPU\state\ (0.6.5), so the check will expect a
clean uninstall and is updated with that change, not here.

What stays: test_re9 expects the unrecorded mirror copy to come back, and
test_edition_download lives in launcher_edition_cases.py (the suite file is
at 262132 bytes on main; the test_re9 text alone would pass the 256 KB
limit without it).

Verified: the file parses; the moved section's body is byte-identical to
main's; check_source_size.py reports 256799 bytes; reuse lint passes. The
test_re9 check itself is unchanged since it was run under Wine against
main's launcher (it passed there); not rerun after this merge.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Signed-off-by: Jonathan Shin <85975340+j0nsh1n@users.noreply.github.com>
@j0nsh1n j0nsh1n changed the title Launcher suite: the two uninstall checks of #219 follow the file-ownership rule Launcher suite: test_re9's uninstall check follows the file-ownership rule (#219) Nov 15, 2026
@j0nsh1n

j0nsh1n commented Nov 15, 2026

Copy link
Copy Markdown
Contributor Author

Done: the PR now carries only the test_re9 change. test_rr and rr.h's header are back to main's text, since both change again with the WrongGPU\state\ design.

Two notes:

  • test_edition_download still moves to launcher_edition_cases.py. run_launcher_tests.py is at 262132 bytes on main, and the test_re9 text alone would pass the 256 KB limit. The moved body is byte-identical. Say if you'd rather make that room another way.
  • Merged with main (after Launcher: the MinGW build of WrongGPU and the suite's --only (#219) #227). The test_re9 check is unchanged since I ran it under Wine against main's launcher; I did not rerun it after this merge.

@phantoslayer21 phantoslayer21 left a comment

Copy link
Copy Markdown
Owner

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looks right. The test_re9 expectation now matches the ownership rule: the test drops the mirror copy from the record itself, so it comes back as the player's. Moving test_edition_download to launcher_edition_cases.py to stay under the size limit is fine. Merging once CI is green. A Windows run of the suite on main follows with the #224 work.

@phantoslayer21
phantoslayer21 merged commit c063935 into phantoslayer21:main Nov 15, 2026
9 checks passed
j0nsh1n added a commit to j0nsh1n/wronggpu that referenced this pull request Nov 15, 2026
…21#226)

One conflict, in run_launcher_tests.py: main (phantoslayer21#222) keeps a two-line
test_edition_download that calls launcher_edition_cases.py; the branch had
called the module from main(). Main's form is kept, and the Linux run skips
that section through it.

phantoslayer21#222 put test_re9's corrected check on main, so its entry leaves
native_linux.KNOWN_ON_MAIN: the check runs and passes natively. The
test_rr check stays skipped until main's fix for phantoslayer21#219 lands.

Verified: fresh build, ctest 22 of 22; launcher-native 366 passed, 0
failed, 5 skipped; reuse lint passes.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Signed-off-by: Jonathan Shin <85975340+j0nsh1n@users.noreply.github.com>
@j0nsh1n
j0nsh1n deleted the test/suite-ownership-checks branch November 15, 2026 05:01
j0nsh1n added a commit to j0nsh1n/wronggpu that referenced this pull request Nov 15, 2026
…21#226)

One conflict, in run_launcher_tests.py: main (phantoslayer21#222) keeps a two-line
test_edition_download that calls launcher_edition_cases.py; the branch had
called the module from main(). Main's form is kept, and the Linux run skips
that section through it.

phantoslayer21#222 put test_re9's corrected check on main, so its entry leaves
native_linux.KNOWN_ON_MAIN: the check runs and passes natively. The
test_rr check stays skipped until main's fix for phantoslayer21#219 lands.

Verified: fresh build, ctest 22 of 22; launcher-native 366 passed, 0
failed, 5 skipped; reuse lint passes.

Signed-off-by: Jonathan Shin <85975340+j0nsh1n@users.noreply.github.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants