Known issues — running notes for zsp

Tracking document for problems found while working across the projects under the Zen Spectrum Project umbrella. Not published anywhere yet — this is the holding pen until the zsp site exists, at which point each entry below moves to that project's page (fixed items as changelog/notes, open items as known limitations).

Status markers: ✓ fixed · ☐ open/flagged · ☑ resolved by decision (e.g. removal)

zentools

No issues found so far (build ✓, vet ✓, test ✓, HEAD fb679a2, v0.5.0).

zen80

☐ ZEXALL/ZEXDOC exhaustive exercisers cannot be verified in a constrained environment. TestZEXALL_CPMSim_BDOS_PCTrap and TestZEX_CPMSim_BDOS_PCTrap run for several minutes on real completion. Their own "silent loop" bailout thresholds (and the dedicated TestZEXALL_QuickCheck's 100M-op budget) are set below what even the first exhaustive subtest (<adc,sbc> hl,<bc,de,hl,sp>....) needs to produce output, so any run bounded to sandbox-feasible time reports a stall. This is a test-harness/environment limitation, not evidence of a Z80 core defect — but it also means the core hasn't actually been confirmed correct by these suites in any session so far. Needs a full local run to completion to confirm one way or the other. (HEAD bd22cfb, v0.1.0+1)

zenzx

✓ Fixed — snapshot retrieval bug (released, v0.4.2). Snapshots saved after loading a screen directly (e.g. a .scr) captured a blank screen: the border restored correctly but the bitmap didn't, because toMachineState encoded stale RAM instead of the authoritative display buffers. Fixed by copying the buffers into the correct bank before encoding, mirroring resyncAfterLoad. Already released as part of v0.4.2 — found by Horacio while prepping that release.

✓ Fixed — unkeyed raylib struct literals (session fix, not yet released). go vet flagged unkeyed rl.Rectangle/rl.Vector2 literals in display.go. Keyed them to match the style already used elsewhere in the same function. Cosmetic/hygiene only, no behavioural change. (HEAD f84a239 + local fix, v0.4.2)

zenas

No issues found so far (build ✓, vet ✓, test ✓, HEAD 97d0006, v0.7.5).

zenimate

✓ Fixed — zaniplay wouldn't compile (session fix, not yet released). cmd/zaniplay/main.go indexed the bit-packed Frame ([]byte) as if it were a plain []bool. Frame already has a purpose-built accessor, At(x, y, w int) bool. The same mistake was duplicated in two test files (cmd/zaniplay/load_test.go, cmd/zenimate-tui/keys_test.go). All three fixed to use .At(...); build/vet/test green afterwards.

☑ zaniplay slated for removal in a future version. zaniplay is absent from both .github/workflows/ci.yml (build/vet/test scoped to pkg/..., internal/..., cmd/zenimate-gui/ only) and release.yml (only zenimate-gui is built for release). Horacio has confirmed the intent is to delete cmd/zaniplay/ entirely in the next version, plus a CHANGELOG note recording the removal.

☐ Open question — cmd/zenimate-tui has the identical CI gap. Same exclusion pattern as zaniplay, same two workflow files. Not yet decided whether this is heading the same way or simply hasn't been promoted into CI. Flagged, awaiting a decision. (HEAD 6e8cf43 + local fixes, v0.7.0)

plus3

No issues found so far (build ✓, vet ✓, test ✓, HEAD 546cede, v0.9.8).

bdf-fonts

☐ Likely accidental duplicate file. fonts/helvR12 copy.bdf sits alongside fonts/helvR12.bdf — the space in the filename is the usual signature of an OS-level duplicate rather than an intentional variant. Not touched; needs a human decision on whether to delete it. (HEAD 1d6ab5c, not a Go module — font/asset repo)

Log