feat(home): a Pull request track builds a Longshot pull request, with one warning for a BigBean's or an approved one and two for any other #199
No reviewers
Labels
No labels
area/build
area/install
area/saves
area/tooling
area/ui
area/updates
ci: approved
good first issue
gpu/amd
gpu/intel
gpu/nvidia
help wanted
kind
bug
kind
chore
kind
docs
kind
feature
kind
performance
kind
question
platform/linux
platform/steam-deck
platform/windows
priority
critical
priority
high
priority
low
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
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set
Reference
MagicBeanDev/LongshotLauncher!199
Loading…
Reference in a new issue
No description provided.
Delete branch "HumanBean/LongshotLauncher:feat/pr-track"
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?
To try a Longshot pull request today, a tester has to check it out and build it from source, so most of the hardware we need never runs a change before it is merged. This adds a Pull request track to the home. It builds an open pull request on the tester's PC the way Beanstalk's agents do. It warns once for a BigBean's or a Beanstalk-approved pull request, and twice for any other.
It rebuilds #59 on today's home (the Track and Version dropdowns), and #59 can be closed in favour of it.
Requires: Beanstalk#90 (
GET /api/v1/public/pulls). It is merged and deployed: production runs coordinator image 51, at e4316243.What changed
Track: Pull request (
HomeViewModel,VersionPick.OfPullRequests). The Version list shows:Drafts (Forgejo's mark or
WIP:) and pull requests labelled for the other system are left out.Switching: picking another pull request shows Switch to #N. It builds in the installed one's place, in its folder (
builds/pr/pr-<number>-<commit>, renamed). Build another version still builds one beside it.Updates: a pull request's only update is a newer push of the same pull request (
UpdatePolicy). Opposite the title it says "#1548 has a newer push: a7c4410e", with Read it in place of What's new. Once the pull request is merged or closed, its row says so.Warnings (
PullRequestViewModels.cs). They come before any build of new pull request code: a first build, a switch, or an update to a new push. A rebuild of a push already built skips them.No load on git.magicbean.dev (
Core/PullRequests.cs):The build (
PullRequestSource,Toolchains.InstallPullRequest). It is Beanstalk's recipe: the PR's source overmain's tools, which here means the newest nightly's.archive/<full hash>.tar.gz, and its pax header must name that commit. The archive is plain files only, with nothing outside its folder.tools.lockdiffers from the nightly's is refused.Builds stay fast. These were already in place, and this keeps them working for pull requests:
nx2/alone takes the nightly's translation instead of translating again.A bigger window: 1040×680, up from 960×600 (
MainWindow.axaml). It still fits a Steam Deck (1280×800) and a 1366×768 laptop, and every page stretches to it.Docs:
docs/pull-requests.mdcovers the flow, the warnings, where the facts come from and the risk. I didn't touchdocs/releases.md, since it is shared with Longshot's repository. The README's feature table links the new page.Tests:
tests/launcher_core/PullRequestChecks.cscovers:../;InstallPullRequest: source replaces, removed files gone, helpers copied, record trimmed, nightly untouched and deletable, a differenttools.lockrefused and leaving nothing;tests/test_launcher_core.py:pr-entries in the track and unordered vectors.home-pr-track,home-pr-versions,home-pr-switch,pr-warning-trusted,pr-warning,pr-confirm,pr-find.Screenshots
Testing
Tested (Linux, .NET 10):
dotnet build LongshotLauncher.slnx: 0 warnings, 0 errors.python -I tests/test_launcher_core.py: 2247 checks pass, with one exception. The existingBuildReportChecks"home folder taken out before a line is cut" check fails on this PC, and it fails the same way on a cleanmain(67ee5f82). With it skipped locally, everything after it passes; the check itself is unchanged.python -I tests/test_release.py: OK.scripts/translations.py: no new faults. The new texts show in English, like the 230-odd already missing from every catalog.tools/screenshots.b268d901, 20 MB) has the commit in its pax header and only files and folders.Not tested:
Risk and rollback
BuildRegistry.GonetreatsC:\…as a deleted folder. The new scenes use Linux paths, and that bug is left for another PR.🤖 Generated with Claude Code
Screenshots (tools/screenshots):
88e760f181ac12faca3fThe window is now 1040×680, up from 960×600. Screenshots retaken at the new size:
WIP: feat(home): a Pull request track builds a Longshot pull request, with one warning for a BigBean's or an approved one and two for any otherto feat(home): a Pull request track builds a Longshot pull request, with one warning for a BigBean's or an approved one and two for any other