Make GRAPHIC select the text plane (#23) #29
Reference in New Issue
Block a user
Delete Branch "23"
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?
Closes #23.
Implements C128-style full-bitmap and split text/graphics composition through the optional TextSink graphic backend hook. GRAPHIC 0 restores the full retained text region; GRAPHIC 1 hides it; GRAPHIC 2 selects the first visible text row. The screenshot tool, graphics guide, embedding API excerpt, tee forwarding, and generated screenshot are updated.
Validation: cmake --build build-akgl --parallel 2; ctest --test-dir build-akgl --output-on-failure -j 2 (111 passed, 1 intentionally skipped interactive typing test); docs_examples verified 96 programs and 19 figures.
@logikoma the tests are failing
@tachikoma can you review this please and leave comments
Beginning CI failure investigation. I have confirmed all five jobs fail on head 3342f2b56995397a5763bfeb2018e5e26492abcb; I am reproducing the failure locally and will post the cause and correction.
Investigation update: the default configuration reproduces cleanly on current head (configure/build succeeded; 112/112 tests passed). Forge commit status still reports five failed CI jobs: cmake_build, sanitizers, coverage, akgl_build, and mutation_test. The actions API rejects non-owner job-log access, so I cannot identify the runner-side failure from available telemetry. Per monitoring scope, I have made no branch changes.
@logikoma they're all failing because of the docs_examples test
CI is red on this PR, and it is not this branch.
No job has passed on any run since run 9 (
812d846, 2026-08-01), acrossmainand every branch pushed since. The last run with greencmake_build,sanitizersandmutation_testwas run 8 (23ccb66, 2026-07-31).akgl_buildandcoverageare worse than that: they have failed on every run in which they appear, including runs 2 through 8, where the other three jobs were green. So two of the five jobs have never passed at all.Reproduced this branch (
3342f2b) locally against the same steps the workflow runs:cmake -S . -B build && cmake --build build --parallel— clean, no warnings.ctest --test-dir build --output-on-failure— 112/112 passed.cmake -S . -B build-asan -DAKBASIC_SANITIZE=ON -DCMAKE_BUILD_TYPE=Debug, thenctest --test-dir build-asan --output-on-failure— 112/112 passed, no ASan or UBSan reports.cmake_buildandsanitizerstherefore fail in CI on inputs that pass on a developer machine, on a branch whose parent commit was already failing them both. The red X on this PR carries no information about this change.I could not retrieve the runner logs to name the cause.
GET /api/v1/repos/andrew/akbasic/actions/runs/{id}/jobsreturns{"jobs":[],"total_count":0}for every run and job id tried, and.../actions/jobs/{id}/logsreturns 404, so the failing step is visible only in the web UI. One candidate is ruled out by direct test: the pinned submodule commits fordeps/libakgl(bbb7b8f) anddeps/basicinterpret(d76162c) both fetch anonymously, soactions/checkout@v4withsubmodules: trueis not failing on an unreachable submodule.This wants its own issue against CI rather than a change to this branch.
Tachikoma (Claude Code Opus 5 1M)
@logikoma please create a new issue against this repository capturing the issues Tachikoma desribed regarding the failing CI build and link it here
Created #30 to track the CI runner/workflow failures described in the review. It records the cross-branch failure history, successful local reproductions, submodule check, and unavailable job-log diagnostics.