feat(evidence): retain paragraph-span prototype as superseded evidence - #56
feat(evidence): retain paragraph-span prototype as superseded evidence#56seonghobae wants to merge 5 commits into
Conversation
Blank-line units keep source positions for later meaning search. Collapsing a multi-paragraph document into one bag-of-words unit fails closed. No new migration.
|
Important Review skippedAuto reviews are disabled on base/target branches other than the default branch. Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Plus Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
|
Superseded by #83. This branch correctly preserves exact paragraph offsets, but it treats PR #83 proposes ADR 0017, a PRD v0.5 delta, research traceability, a full design, and an atomic TDD implementation plan. It preserves paragraph boundaries as one source-structure signal rather than accepting them as semantic truth. Closing this draft avoids parallel, contradictory public APIs and duplicate Checks. No implementation or release claim is transferred automatically; #83 remains a proposed design pending exact-head checks and independent review. |
Superseded by PR #83
This branch established useful exact-offset evidence, but its public semantic-unit contract equates units with blank-line paragraph splitting. That is not sufficient authority for mixed scripts, headings, lists, tables, code, captions, DOM structure, one oversized paragraph, exact final-payload token budgets, recursive recovery, hierarchy or context restoration.
PR #83 is the canonical design owner for the same TEPP bounded context. It preserves the useful exact-source-span direction while adding language-agnostic structure, immutable model profiles, rendered-payload accounting, lossless oversized-unit recovery, hierarchy,
embedding_space_id, provider-neutral boundaries and APA 7 research traceability.Preserve this branch and its regressions as historical implementation evidence. Port any unique useful tests into the post-#83 implementation plan rather than maintaining two competing semantic-unit contracts.
Closing this PR prevents ambiguous ownership and duplicate implementation lanes; it does not claim that PR #83 has already implemented the production capability.
Superseded by #83.