qt: build and test on every platform it can support #4554

Activity

1. TrapsterDK

   added

   needs-hostNeeds macOS, Windows, GPU, or signing secrets; not doable on LinuxNeeds macOS, Windows, GPU, or signing secrets; not doable on Linux

   and removed

   needs-hostNeeds macOS, Windows, GPU, or signing secrets; not doable on LinuxNeeds macOS, Windows, GPU, or signing secrets; not doable on Linux

   on Oct 28, 2026
2. TrapsterDK commented on Oct 28, 2026

   

   TrapsterDK

   on Oct 28, 2026

   CollaboratorAuthor

   More actions

   Maintainer policy update: everything should build and test on every platform it supports. Add the CI coverage for the supported platforms now; do not wait for an evidence dispatch or CI before widening the constraint. If the language or target genuinely cannot run on a platform, mark that platform n/a in the support matrix with a one-line reason (see #4666) instead of leaving it as a gap.
3. TrapsterDK commented on Oct 28, 2026

   

   TrapsterDK

   on Oct 28, 2026

   CollaboratorAuthor

   More actions

   macos_arm64 now reported via the PR above; Windows cells remain blocked on #3302.
4. TrapsterDK

   added a commit that references this issue on Oct 28, 2026

   qt: report qt_test on every qualified cell in the support matrix (Refs …

   ...

   a26f994
5. TrapsterDK

   added

   blockedWaiting on upstream or a design decisionWaiting on upstream or a design decision

   on Oct 29, 2026
6. TrapsterDK commented on Oct 29, 2026

   

   TrapsterDK

   on Oct 29, 2026

   CollaboratorAuthor

   More actions

   Triage: Issue #4554 (qt build gaps, 3 cells) is partly done. macos_arm64 closed by #4715, and GAP_ISSUES now only lists build:qt:windows_*. The two Windows cells remain: windows_x86_64 is blocked in pins.bzl and windows_arm64 has no cell. Both depend on #3302 (pinned MSVC Qt 6.11.3 archives, Windows extractor, DLL layout, platform entries). That is a new SDK subsystem, so size is L and I marked it risky because it touches launcher code. The issue body forbids n/a marks, but a later comment allows n/a for real impossibility. Qt ships Windows builds, so n/a does not fit. Issue is blocked on #3302.
7. TrapsterDK

   added

   model:humanNeeds real hosts, secrets, releases, upstream PRs or an owner decisionNeeds real hosts, secrets, releases, upstream PRs or an owner decision

   on Nov 1, 2026
8. TrapsterDK

   changed the title [-]Support matrix gaps: build qt (3)[/-] [+]qt: build and test on every platform it can support[/+] on Nov 2, 2026
9. TrapsterDK

   added

   P3New languages, platforms and features, lastNew languages, platforms and features, last

   model:opusAgent lane: needs planning or open design, OpusAgent lane: needs planning or open design, Opus

   and removed

   model:humanNeeds real hosts, secrets, releases, upstream PRs or an owner decisionNeeds real hosts, secrets, releases, upstream PRs or an owner decision

   on Nov 2, 2026
10. TrapsterDK commented on Nov 4, 2026

    

    TrapsterDK

    on Nov 4, 2026

    CollaboratorAuthor

    More actions

    Plan (agent): size L, split into existing issues (no new ones)
    Decision: make this a tracking issue over the slices that already exist under #3302 (SDK cells) and #5196 (rules and bundle). Do not file duplicates. Rejected: new slices here (they would copy #5104/#5109/#6519/#6520/#6521 word for word); marking Windows n/a (Qt ships MSVC x64 and arm64 builds, so it is not impossible); keeping this blocked (nothing needs the owner, only Windows evidence runs).
    Cause: qt/sdk/windows_x86_64.bzl on origin/main bbac89b has IDENTITY = None, needs = None and blocker = "MSVC toolchain and Windows probe evidence pending (#4815, #5972)", so qualified_cells() skips it. windows_arm64 is the same kind of placeholder. tools/docs/support_matrix/languages.bzl:52 derives the qt cells from qualified_cells().
    Note: GAP_ISSUES no longer exists on main. The matrix cells close on their own once the cells qualify, so there is no entry to delete.
    Order:

    1. qt: Qt rules match Qt libraries by stem and drop rpath and -fPIC on Windows #6519: Qt rules match libraries by stem; no rpath and no -fPIC on Windows (S, Linux only, can start now)
    2. cc: implement and qualify the Microsoft cl.exe Windows x64 toolchain #4815: MSVC cl.exe toolchain qualified on Windows x64 (needs-host; qt: the pinned Qt SDK is acquired on Windows x86_64 #5104 needs it)
    3. qt: the pinned Qt SDK is acquired on Windows x86_64 #5104: pinned Qt SDK acquired on windows_x86_64 (needs-host)
    4. qt: qt_bundle and the Qt tests run on windows_x86_64 #6520: qt_bundle and Qt tests on windows_x86_64 (M, needs-host, after qt: Qt rules match Qt libraries by stem and drop rpath and -fPIC on Windows #6519 and qt: the pinned Qt SDK is acquired on Windows x86_64 #5104)
    5. qt: the pinned Qt SDK is acquired on Windows arm64 #5109: pinned Qt SDK acquired on windows_arm64 (needs-host, after qt: the pinned Qt SDK is acquired on Windows x86_64 #5104)
    6. qt: qt_bundle and the Qt tests run on windows_arm64 #6521: qt_bundle and Qt tests on windows_arm64 (S, needs-host, after qt: qt_bundle and the Qt tests run on windows_x86_64 #6520 and qt: the pinned Qt SDK is acquired on Windows arm64 #5109)
       Tests: one per slice, as set in each issue. The final proof is gh workflow run evidence.yml -f scope=//qt/... -f platforms='["windows_x86_64","windows_arm64"]' with //qt/tests:qt_test, //qt/tests:gui_test and //qt/tests:qt_app_consumer_test showing PASSED, not SKIPPED.
       Host: windows_x86_64 and windows_arm64 evidence runs (in the slices).
       Upstream: none.
       Close this issue when qt: acquire the pinned Qt SDK on Windows x86_64 and Windows arm64 #3302 and qt: Qt rules and qt_bundle support the Windows cells #5196 are closed.
11. TrapsterDK

    added

    trackingParent issue split into sub-issuesParent issue split into sub-issues

    and removed

    blockedWaiting on upstream or a design decisionWaiting on upstream or a design decision

    on Nov 4, 2026
12. TrapsterDK

    added a commit that references this issue on Nov 5, 2026

    qt/sdk: windows_x86_64 blocker cites open issues (part of #4554) (#6936)

    ...

    0a10e99
13. TrapsterDK

    added

    blockedWaiting on upstream or a design decisionWaiting on upstream or a design decision

    and removed

    model:opusAgent lane: needs planning or open design, OpusAgent lane: needs planning or open design, Opus

    on Nov 5, 2026
14. TrapsterDK commented on Nov 5, 2026

    

    TrapsterDK

    on Nov 5, 2026

    CollaboratorAuthor

    More actions

    Blocked: remaining Windows Qt cells wait on open #4815/#6827 (MSVC cl.exe payload failure, evidence runs 42005669446 and 42006944062), which gate #6520, #5109 and #6521. No code slice can be proven until cl.exe works on Windows.

    Assisted by Claude.
15. github-actions commented on Nov 6, 2026

    

    github-actionsBot

    on Nov 6, 2026 – with GitHub Actions

    More actions

    This issue has the blocked label but no Blocked on: line in its body. Add one naming the blocking issue (#N), a pull request URL, or the owner action. See CONTRIBUTING.md.
16. TrapsterDK

    removed

    blockedWaiting on upstream or a design decisionWaiting on upstream or a design decision

    on Nov 7, 2026
17. TrapsterDK

    added a commit that references this issue on Nov 7, 2026

    qt: README records the windows_arm64 msvc tests in CI (#4554)

    ...

    7c70081
18. TrapsterDK commented on Nov 7, 2026

    

    TrapsterDK

    on Nov 7, 2026

    CollaboratorAuthor

    More actions

    Evidence run 42385700174: //qt/tests:qt_test_msvc and //qt/tests:gui_test_msvc PASSED on windows_x86_64 and windows_arm64. PR #9353 closes this.