Brad's ruling (2026-09-18 01:15): "I want to add a follow-up to compare all of our effects to all of hexeguitar's IF we find hexeguitar's work is worthy in Phase 5. I don't want to go back and change anything in phases up through 4 just for the sake of changing. But if we find hexeguitar's algorithms superior in an effect, we should use it."
Trigger: Phase 5 (Time) builds against his spring reverb, plate, ReverbSc FDN and modulated stereo delay as design references (hexefx_audiolib_F32 / _i16, MIT per file for those four — see the survey), and the Phase 5 gates say whether they are worthy: traits held over the span, cost inside the palette budget on the P4 and the S3, three interpreters identical, alias and noise floors. If they are, this follow-up opens.
Scope when it opens: every effect class of ours with a counterpart in his libraries (phasers incl. the infinite one, delay, reverbs, compressor, noise gate, booster/overdrive, wah, tone stack, cabinet sim, 3-band EQ, DC blocker), compared on our gates, not by taste. Where his is measurably superior — a trait we could not hold that his holds, a lower cost at equal traits, a cleaner floor — the class adopts his algorithm through the normal chain (build → refute → audit → boards), with licence per file respected: spring, plate, delay are MIT; tone stack and wah are GPL-3.0 and can only inform a design, never be ported; ReverbSc's Csound lineage is checked first. Where ours is as good, nothing changes.
Survey: anchor docs/effects-internal/dossiers/sources-hexefx.md. Phase 5 tracking: #33.
Brad's ruling (2026-09-18 01:15): "I want to add a follow-up to compare all of our effects to all of hexeguitar's IF we find hexeguitar's work is worthy in Phase 5. I don't want to go back and change anything in phases up through 4 just for the sake of changing. But if we find hexeguitar's algorithms superior in an effect, we should use it."
Trigger: Phase 5 (Time) builds against his spring reverb, plate, ReverbSc FDN and modulated stereo delay as design references (
hexefx_audiolib_F32/_i16, MIT per file for those four — see the survey), and the Phase 5 gates say whether they are worthy: traits held over the span, cost inside the palette budget on the P4 and the S3, three interpreters identical, alias and noise floors. If they are, this follow-up opens.Scope when it opens: every effect class of ours with a counterpart in his libraries (phasers incl. the infinite one, delay, reverbs, compressor, noise gate, booster/overdrive, wah, tone stack, cabinet sim, 3-band EQ, DC blocker), compared on our gates, not by taste. Where his is measurably superior — a trait we could not hold that his holds, a lower cost at equal traits, a cleaner floor — the class adopts his algorithm through the normal chain (build → refute → audit → boards), with licence per file respected: spring, plate, delay are MIT; tone stack and wah are GPL-3.0 and can only inform a design, never be ported; ReverbSc's Csound lineage is checked first. Where ours is as good, nothing changes.
Survey: anchor
docs/effects-internal/dossiers/sources-hexefx.md. Phase 5 tracking: #33.