Conversation
Investigated whether the '*'-traversal/GROUP-mode content-accessor rejection (FILES) was safe to convert to the deferred ambiguous_content_read pattern the same way this session's other three field-accessor fixes were, without a check-in first, since none of them needed to. It is not a safe blind conversion: the existing ambiguous_content_read flag is a naive count check that does not know whether candidates were already individually reduced one-per-partition (legitimate, per this bucket-list item's own CSVPATHS precedent) versus genuinely unreduced and sharing one accessor (a real Rule 1 violation). Field accessors never trip this flag at all, which is why narrowing to field-accessors- only was safe without touching it. Making :manifest() safe here too needs the flag itself to become partition-aware first. Documented the exact fix direction, likely affected call sites (RESULTS' own equivalent GROUP-mode check, possibly its literal-root match_all check), and the scope decision needed before building (FILES-only vs. also RESULTS/CSVPATHS in one pass) -- no code changed. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013gPjejPWvbWtjz2dw9h14M
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
No code changes -- documentation only. Investigated whether the
'*'-traversal/GROUP-mode content-accessor rejection (FILES) was safe to
convert to the deferred
ambiguous_content_readpattern, the same waythis session's other three field-accessor fixes were (#274, #275, plus
the RESULTS narrowing fix in #273), without a check-in first -- since
none of them needed one.
It is not a safe blind conversion. Wrote up the exact reasoning and the
fix direction in
deferred_work_bucket_list.mdfor review rather thanbuild it unreviewed while unavailable overnight, per the bucket list's
own existing warning on this item ("not obviously safe to convert...
not a blind port").
What's documented
ambiguous_content_readflag is a naive count check,unsafe to rely on for
:manifest()once GROUP-mode + pointer isallowed through (it cannot currently tell "several groups, each
already reduced to one via its own pointer" apart from "several raw,
unreduced candidates").
has_manifest, so the flag staysFalseregardless of count) --confirming the three field-accessor PRs already merged are not at
risk.
reasoning CSVPATHS' own already-shipped equivalent fix used.
accessor check, possibly its literal-root
match_allcheck).RESULTS/CSVPATHS in one pass.
Test plan
N/A -- no code changed.
https://claude.ai/code/session_013gPjejPWvbWtjz2dw9h14M