Remove the extracted verisimdb tree - #64
Conversation
Closes the second and larger of the two duplications. 1058 -> 346 tracked files. verisimdb/ is now a pointer README. Safe only because the recovery landed first: hyperpolymath/verisimdb#219 is merged, and all eight files it carried are confirmed live on that repo's main before anything was deleted here. == This was a fork, not a duplicate lithoglyph/ was removed after verifying every file byte-identical to its own repo. That check FAILED here: only in this directory 108 only upstream 175 common 608, of which 333 differ (331 substantive) Deleting on the lithoglyph precedent would have destroyed ~4,000 lines including an entire tier of test coverage. The transferable lesson is the check, not the deletion. == Recovered before removal (verisimdb#219) Seven files had no upstream counterpart of any kind — not moved, not renamed, not converted — and none was a stub: concurrency_test.exs 444 security_test.exs 436 e2e_verisimdb_test.exs 410 kraft_property_test.exs 328 groove.rs 890 a2ml.rs 360 ram_promotion.rs 472 Plus WHITEPAPER.pdf. The four Elixir suites run against upstream unmodified (5 properties, 43 tests, 0 failures); the Rust modules are declared in their crates' lib.rs and compile (cargo check exit 0), verified by negative control. storage_regenerator.rs (682 lines) was deliberately NOT ported: upstream already has verisim-normalizer/src/regeneration.rs and verisim-api/src/regenerator.rs. Superseded, not lost. The other 100 were triaged as moved (16), converted (26), or superseded prose, config and retired ReScript. == Consequence for the anti-pattern gate .res files: 44 -> 4 Only typeql-experimental/ remains. The gate has been red on ReScript since before this branch; it is now within reach of green for the first time, on the last legacy directory that still carries any. .ts/.tsx 0, package.json 0 — unchanged, both already clear. == Guards and docs updated placement-guard: verisimdb removed from GRANDFATHER, so a new file under verisimdb/ now FAILS rather than warns — same treatment lithoglyph/ got on 2026-07-27. Warning would let the duplication grow back. CLAUDE.md and REGISTRY.adoc: both extractions recorded as done. Two stale REGISTRY rows corrected while there — Lithoglyph and Glyphbase were still listed "to create" when both have existed for over a week, and Glyphbase was described as the "web UI" rather than Airtable-mode delivery. Both files now carry the warning that verisimdb was a fork, so the next extraction is not assumed to be a clean subset. == Nothing is lost $ git ls-tree -r split-history/verisimdb --name-only | wc -l 713 Branch _split_verisimdb and tag split-history/verisimdb, both on origin, 124 commits. Never prune them, nor their lithoglyph/glyphbase/gnpl equivalents. a2ml and k9 validators pass; all workflows parse.
|
Note Automatic reviews are paused because your trial's included automatic processing has been used for this period. Upgrade now, or comment "Gitar review" to run a review anytime. Code Review ✅ ApprovedRemoves the extracted verisimdb tree and replaces it with a pointer README, reducing tracked files from 1058 to 346. No issues found.
OptionsDisplay: compact → Showing less information. Comment with these commands to change the behavior for this request:
Important Your trial ends in 7 days — upgrade now to keep code review, CI analysis, auto-apply, custom automations, and more. Was this helpful? React with 👍 / 👎 | Gitar |
|
|
Closes the second and larger of the two duplications. 1058 → 346 tracked files.
verisimdb/is now a pointer README.Safe only because the recovery landed first:
verisimdb#219is merged, and all eight files it carried were confirmed live on that repo'smainbefore anything was deleted here.This was a fork, not a duplicate
lithoglyph/was removed after verifying every file byte-identical to its own repo. That check failed here:Deleting on the lithoglyph precedent would have destroyed ~4,000 lines including an entire tier of test coverage. The transferable lesson is the check, not the deletion.
Recovered before removal
Seven files had no upstream counterpart of any kind — not moved, not renamed, not converted — and none was a stub:
concurrency_test.exssecurity_test.exse2e_verisimdb_test.exskraft_property_test.exsgroove.rsa2ml.rsram_promotion.rsPlus
WHITEPAPER.pdf. The four Elixir suites run against upstream unmodified — 5 properties, 43 tests, 0 failures — and the Rust modules are declared in their crates'lib.rsand compile (cargo checkexit 0), verified by negative control.storage_regenerator.rs(682 lines) was deliberately not ported: upstream already hasverisim-normalizer/src/regeneration.rsandverisim-api/src/regenerator.rs. Superseded, not lost.The other 100 were triaged as moved (16), converted (26), or superseded prose, config and retired ReScript.
The anti-pattern gate is now within reach
Only
typeql-experimental/still carries any. That gate has been red on ReScript since before this branch; this is the first time green is achievable..ts/.tsxandpackage.jsonwere already at 0.Guards and docs
placement-guard:verisimdbremoved fromGRANDFATHER, so a new file underverisimdb/now fails rather than warns — the same treatmentlithoglyph/got. Warning would let the duplication grow back.CLAUDE.mdandREGISTRY.adoc: both extractions recorded as done. Two staleREGISTRYrows corrected while there — Lithoglyph and Glyphbase were still listed "to create" when both have existed for over a week, and Glyphbase was described as the "web UI" rather than Airtable-mode delivery.Both now carry the warning that verisimdb was a fork, so the next extraction isn't assumed to be a clean subset.
Nothing is lost
Branch⚠️ Never prune them, nor their
_split_verisimdband tagsplit-history/verisimdb, both on origin, 124 commits.lithoglyph/glyphbase/gnplequivalents.a2ml and k9 validators pass; all workflows parse.
🤖 Generated with Claude Code