Skip to content

Bump the DspTap pin to 28a34a1 - #25

Merged
tap merged 1 commit into
mainfrom
claude/repository-inconsistencies-nkocvd
Jul 28, 2026
Merged

Bump the DspTap pin to 28a34a1#25
tap merged 1 commit into
mainfrom
claude/repository-inconsistencies-nkocvd

Conversation

@tap

@tap tap commented Jul 28, 2026

Copy link
Copy Markdown
Owner

What this changes

Refreshes submodules/dsptap from b3e9ee5 to 28a34a1 (current DspTap main) — eight commits.

Four are infrastructure (TapHouse drift pin, SessionStart hook, CLAUDE.md, PR template). The code-bearing ones add the shared FIR substrate carried from SampleRateTap (kaiser.h, sample_traits.h, fir_kernels.h, quantize.h, analysis/) and dot_row_reversed, plus a sign-conversion fix in DspTap's own kaiser test.

Why

The pin was the second-stalest in the family. Nothing here calls the new headers — this repo reaches into tap::dsp for the real FFT (stft.h, tune.h, conv_engine.h) and the pitch primitives, all unchanged across the range — so the bump is functionally a no-op that keeps the pin current. That matters because TapTools-Max picks this up transitively through its own submodules/taptools pin, so letting it drift here compounds downstream.

Verification

Configure and build clean, zero errors. The Catch2 kernel suite passes 1,637,526 assertions across 156 test cases.

One thing worth stating so the number isn't misread: ctest reports this as 1 test, because this repo registers a single aggregate add_test rather than discovering cases individually. "1/1 passed" is the entire battery, not one check.

Checked before bumping, so the low risk is established rather than assumed:

  • every change under DspTap's include/ across the range is an addition — 9 new headers, none modified or deleted
  • exported targets identical at both ends: tap_dsp_fft, tap::dsp_fft, tap_dsp, tap::dsp, tap_dsp_warnings
  • no consumer-facing CMake change — no target, option, find_package or FetchContent diff
  • the one bug fix in the range touched DspTap's own tests/test_kaiser.cpp, not a shipped header

Notes for the reviewer

Pinned to a commit on DspTap's main, not to any in-flight branch — consumer pins must reference a tree that stays reachable after branch cleanup. In particular this is independent of tap/DspTap#8 (which only adds a CMake option and a CI flag), so neither PR blocks the other.

Part of a four-repo sweep: RatioTap, SampleRateTap and AmbiTap get the same bump. MuTap is deliberately excluded — it's 11 commits behind, but its macOS CI is red from tap/MuTap#31, and a dsptap bump is one of the candidate fixes listed in that issue. Bumping it here would tangle a routine refresh with an experiment, and make a still-red CI ambiguous.


Generated by Claude Code

Refreshes submodules/dsptap from b3e9ee5 to 28a34a1, the current DspTap main --
eight commits behind. Four are infrastructure (TapHouse drift pin, SessionStart
hook, CLAUDE.md, PR template); the code-bearing ones add the shared FIR
substrate carried from SampleRateTap (kaiser.h, sample_traits.h, fir_kernels.h,
quantize.h, analysis/) and dot_row_reversed, plus a sign-conversion fix in
DspTap's own kaiser test.

Nothing here calls the new headers -- this repo reaches into tap::dsp for the
real FFT (stft.h, tune.h, conv_engine.h) and the pitch primitives, which are
unchanged across the range. So the bump is functionally a no-op that keeps the
pin current, which matters because TapTools-Max picks this up transitively
through its own submodules/taptools pin.

Pinned to a commit on DspTap's main rather than any in-flight branch, per the
release flow: consumer pins must reference a tree that stays reachable after
branch cleanup.

Verified: configure and build clean with zero errors; the Catch2 kernel suite
passes 1637526 assertions across 156 test cases. Note ctest reports this as a
single test -- this repo registers one aggregate add_test rather than
discovering cases individually, so "1/1" is the whole battery, not one check.

Checked before bumping: across the range every change under DspTap's include/ is
an ADDITION -- nine new headers, none modified or deleted -- the exported targets
are identical at both ends, and there is no consumer-facing CMake change.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JhhQ93r2E1QTnCx46YfX8j
@tap
tap merged commit c91479c into main Jul 28, 2026
12 checks passed
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