#620's read marks a particle inside a chain run BOUND. The peel stops at it, and credential_run gathers an absorbed particle run into one report on the read's path. Both re-create in the reader what the chain's particle-to-particle merge did before #614 split the run off ahead of the chain (#620's /simplify, altitude review).
Proposal: at the TRAILING site, before read_trailing_run, merge each run of two or more adjacent particles (the chain's leading piece excluded, the merged piece keeping the prefix tag). Then delete the BOUND stop and the gathering.
Measured on a prototype (2026-10-08):
- 80 role moves on a 551,880-name grid, all two adjacent duals behind a period title. For example
Dr. mc mc: today given 'Dr.', middle 'mc', family 'mc'; the prototype reads suffix 'mc mc'.
- 3,720 report-only diffs, because the chain's PARTICLE_OR_GIVEN emitter is gated on
j > k + 1 and a pre-merged run makes j == k + 1. It needs re-keying to "claimed more than one token".
Question: is retiring two reader-side mechanisms worth those moves?
#620's read marks a particle inside a chain run BOUND. The peel stops at it, and
credential_rungathers an absorbed particle run into one report on the read's path. Both re-create in the reader what the chain's particle-to-particle merge did before #614 split the run off ahead of the chain (#620's /simplify, altitude review).Proposal: at the TRAILING site, before
read_trailing_run, merge each run of two or more adjacent particles (the chain'sleadingpiece excluded, the merged piece keeping theprefixtag). Then delete the BOUND stop and the gathering.Measured on a prototype (2026-10-08):
Dr. mc mc: today given 'Dr.', middle 'mc', family 'mc'; the prototype reads suffix 'mc mc'.j > k + 1and a pre-merged run makesj == k + 1. It needs re-keying to "claimed more than one token".Question: is retiring two reader-side mechanisms worth those moves?