Skip to content

condemned.md

Latest commit

 

History

History
436 lines (394 loc) · 39.7 KB

condemned.md

File metadata and controls

436 lines (394 loc) · 39.7 KB
Raw

Condemned material

The catalog of condemned self-material — stale branches, git stash entries, dead code and retired tests that a hygiene / repo_cleanup sweep is 95%-but-not-100% sure is trash. Symmetric to parked.md: parked.md holds work that is paused and will resume; this file holds material that is spent and awaiting elimination, recoverable right up until it is voided.

This is the index; the payload is durable git refs, not markdown. Fragile forms (local unmerged branches, stashes) are first materialised as real commits and pushed under the archive namespace refs/heads/archive/condemned/<name> (a branch prefix — GitHub only accepts pushes to refs/heads/* and refs/tags/*, so a custom refs/archive/* namespace is unpushable; filter with git branch --list 'archive/condemned/*') into PyAutoGut as the attic remote — before the local copy is deleted. Recovery is a checkout. The organ (PyAutoGut) holds and voids; the Brain hygiene conductor drives (decides what to condemn, triggers a sweep), mirroring the Heart ↔ vitals template. See the decision: complete/2026/07/pyautogut-organ-decision.md.

Contents

Lifecycle

  1. Condemn — the hygiene tidy pass files an entry here (async, no synchronous per-item gate). Fragile forms are archived to a durable ref first; merged branches and committed deletions need only a SHA recorded.
  2. Transit — the entry sits with a sweep-after date. Until then it is recoverable (reabsorption): restore the branch/stash from its archive ref.
  3. Void — a batch sweep runs the existing repo_cleanup safety gates against entries past sweep-after and eliminates them; the entry moves out of this file (to a voided log or is deleted).

Entry schema

One ## block per item. Fields:

  • type — branch | stash | file | test
  • locator — the local name/path (e.g. feature/old-thing, stash@{2}, src/legacy/foo.py)
  • confidence — how sure it is trash (e.g. 0.95)
  • reason — why it is condemned
  • merged — yes | no (a merged branch is reachable from main forever → skips the pen; near-zero risk)
  • condemned — date filed (YYYY-MM-DD)
  • sweep-after — earliest date it may be voided (the transit clock)
  • breaks-if-wrong — what is lost if this was a false positive (informs the gate)
  • archive-ref — the durable ref + SHA to recover from (refs/heads/archive/condemned/<name> @ <sha>), or n/a for a merged branch / committed deletion whose bytes live in remote history (record the pre-delete SHA instead)

Recoverability is not uniform

  • Merged branches — reachable from main forever; archive-ref: n/a, a note is enough. The conductor recommends these straight to deletion without staging.
  • Committed code / test deletions — the old bytes live in remote history; record only the pre-delete SHA.
  • Local-only unmerged branches / stashes — exist in one machine's reflog and are gc-pruned. These must be materialised as an archive ref before deletion; a manifest that merely points at a stash is worthless the moment it is dropped.

release-datasets/autolens-regenerable

  • type: file
  • locator: autolens_workspace dataset/ (28 non-allowlisted simulated dirs, PR#272) — cat A+multi: group/{dark_matter_subhalo,operated,simple,simple__no_lens_light,sky_background}, imaging/{dark_matter_subhalo,extra_galaxies,lens_light_asymmetric,light_operated,simple,simple__no_lens_light,sky_background}, interferometer/{extra_galaxies,simple}, point_source/{deblending,simple}, weak/simple, multi/{imaging,interferometer}; + producer-output/guarded: cluster/csv_api_example, imaging/{dark_matter_subhalo_no_lens_light,misc,simulated_lens}, interferometer/{datacube,simulated_lens}, point_source/{simulated_lens,start_here_example}; + simpleold (dead, user-confirmed kill)
  • confidence: 0.98 (0.90 for interferometer/simpleold — dead)
  • reason: force-committed by pre_build's old git add -f (fixed PyAutoBuild#150); each is regenerated on demand by a guarded example script (should_simulate()/not dataset_path.exists() → simulator subprocess) or is scratch script-output with no consumer. simpleold is a stale old dataset with zero references. Removing from tracking so the tree matches .gitignore. #126 leg 3.
  • merged: no
  • condemned: 2026-07-13
  • sweep-after: 2026-08-13
  • breaks-if-wrong: a consumer that does not self-provision would hit missing data at runtime; smoke (9/9) confirms the example entries regenerate post-purge; datacube/ellipse have guarded consumers; producer-output dirs have no consumer
  • archive-ref: n/a — committed deletion; bytes recoverable from remote history at pre-purge SHA 8625a1de (autolens_workspace) via git checkout 8625a1de -- dataset/<dir>

release-datasets/autogalaxy-regenerable

  • type: file
  • locator: autogalaxy_workspace dataset/ (17 non-allowlisted simulated dirs, PR#129) — cat A+multi: imaging/{asymmetric,clumpy,extra_galaxies,operated,sersic_x2,simple,simple__sersic,sky_background}, interferometer/{clumpy,extra_galaxies,simple}, multi/{imaging,interferometer}; + producer-output/guarded: imaging/{ellipse,misc,simulated_galaxy}, interferometer/simulated_galaxy
  • confidence: 0.98
  • reason: as above (#126 leg 3); regenerated on demand by guarded example scripts or scratch script-output.
  • merged: no
  • condemned: 2026-07-13
  • sweep-after: 2026-08-13
  • breaks-if-wrong: as above; smoke (8/8) confirms regeneration post-purge
  • archive-ref: n/a — committed deletion; bytes recoverable from remote history at pre-purge SHA e940f8cd (autogalaxy_workspace)

release-datasets-group-b/all-four

  • type: file
  • locator: bare-dataset/ Group B repos (PyAutoBuild#151) — autofit_workspace (408 files, dataset/example_1d/), HowToFit (408, example_1d/), HowToGalaxy (21, imaging/), HowToLens (27, imaging/). Full dataset/ purge — nothing real was committed.
  • confidence: 0.98
  • reason: force-committed by pre_build's old git add -f (fixed PyAutoBuild#150). All four repos self-provision every dataset via if not dataset_path.exists(): subprocess.run([sys.executable, ".../simulator(s).py"]) (autofit/HowToFit path.exists()+from_json; HowToGalaxy/HowToLens dataset_path.exists()+from_fits). Bare dataset/ intent = nothing committed. #151.
  • merged: no
  • condemned: 2026-07-13
  • sweep-after: 2026-08-13
  • breaks-if-wrong: a non-self-provisioning script would hit missing data; smoke confirms regeneration (autofit 10/10, HowToFit 10/10, HowToGalaxy 4/4, HowToLens 6/6)
  • archive-ref: n/a — committed deletion; bytes recoverable from remote history at pre-purge SHAs autofit_workspace 1254a2fe, HowToFit ccc19584, HowToGalaxy caf1657c, HowToLens 554be8b4

pj011646-wfc3-parity/superseded-branch

  • type: branch
  • locator: feature/pj011646-wfc3-parity (PyAutoReduce, tip bd9806b) — scripts/reduce_pj011646.py (wfc3_ir production reduction) + prototypes/pj011646_parity_fit.py (model-parity fit); PyAutoReduce#25.
  • confidence: 0.9
  • reason: superseded — the human is redoing the PJ011646 WFC3-IR reduction fresh via a new autolens-assistant approach (not the PyAutoReduce prototype path). Branch never pushed / no PR; pixel-parity verdict already captured on #25 (closed superseded). Kept recoverable in case the fresh redo wants the reduction/fit scripts as reference.
  • merged: no
  • condemned: 2026-07-13
  • sweep-after: 2026-10-11
  • breaks-if-wrong: loses the working wfc3_ir reduction script (program 14653, F160W, final_bits=512 workaround) + the model-parity fit harness; recoverable via pyauto-gut recover pj011646-wfc3-parity until voided.
  • archive-ref: refs/heads/archive/condemned/pj011646-wfc3-parity on PyAutoReduce origin (bd9806b)

release-datasets/autolens-regenerable-leg2

  • type: file
  • locator: autolens_workspace dataset/ (7 committed simulated dirs, PR#353) — cluster/simple, imaging/{mass_stellar_dark,double_einstein_ring,extra_and_scaling_galaxies}, group/{mass_stellar_dark,double_einstein_ring,scaling_relation}
  • confidence: 0.98
  • reason: committed simulated data (~17 MB) each regenerated on demand by a guarded consumer running its paired simulator; regeneration proven from a clean tree for all 7 before purge (issue #352). .gitignore allowlist re-includes dropped in the same commit; interferometer/uv_wavelengths deliberately kept (real SMA input, no writer). #126 purge series leg (prior: PR#272).
  • merged: no
  • condemned: 2026-07-27
  • sweep-after: 2026-08-27
  • breaks-if-wrong: loses the exact historical noise realizations of the 7 datasets (regenerated copies use fresh noise); recoverable from remote history via the pre-purge SHA
  • archive-ref: n/a — committed deletion; pre-purge SHA 0bb170c57 VOIDED by the 2026-07-27 leg-7 history rewrite — bytes recoverable ONLY from the local mirror backup ~/Code/PyAutoLabs-backups/autolens_workspace-pre-rewrite-2026-07-27.git

release-datasets/autogalaxy-database-orphans

  • type: file
  • locator: autogalaxy_workspace dataset/database/simple__{0,1,2} + scripts/guides/results/database/simulators/light_sersic_exp__{0,1,2}.py (PR#170)
  • confidence: 0.98
  • reason: orphaned committed simulated data (512 KB) — no commit in repo history ever contained a reader of dataset/database/ (the database guide reads dataset/imaging/*; "database" is its output namespace, matching the autolens sibling guide). The 2026-07-13 allowlist call (#126 Group B "committed-by-design aggregator data") rested on a stale prose line fixed in the same PR. Simulators deleted with their output: unconsumed, and their absence from no_run.yaml made every build re-run them (dataset byte-churn commits 34ddb66d/10a1ee64/e942d360). Dataset-bulk series leg 2 (issue #169).
  • merged: no
  • condemned: 2026-07-27
  • sweep-after: 2026-08-27
  • breaks-if-wrong: loses the exact historical noise realizations + the 3 simulator scripts; both recoverable from remote history at the recover point
  • archive-ref: n/a — committed deletion; SHAs e942d360/5632f6d0 VOIDED by the 2026-07-27 leg-7 history rewrite — bytes (incl. the 3 deleted light_sersic_exp simulators, the one non-regenerable item) recoverable ONLY from ~/Code/PyAutoLabs-backups/autogalaxy_workspace-pre-rewrite-2026-07-27.git

release-datasets/autolens-many-visibilities

  • type: file
  • locator: autolens_workspace dataset/interferometer/many_visibilities (PR#356)
  • confidence: 0.98
  • reason: force-committed in a single 2026-05-14 "pre build" add and never referenced — no script writes or loads it (all 9 name mentions are prose pointing at the many_visibilities_preparation example, which reads interferometer/simple); FITS files are 8,640-byte test-resolution stubs, 1.6 of 1.64 MB is render PNGs. Dataset-bulk series follow-on, user-confirmed kill.
  • merged: no
  • condemned: 2026-07-27
  • sweep-after: 2026-08-27
  • breaks-if-wrong: loses nothing unique — stubs + renders, recoverable from remote history
  • archive-ref: n/a — committed deletion; pre-purge SHA 7f6ba9954 VOIDED by the 2026-07-27 leg-7 history rewrite — bytes recoverable ONLY from ~/Code/PyAutoLabs-backups/autolens_workspace-pre-rewrite-2026-07-27.git

history-rewrite-2026-07-27-addendum

  • type: file
  • locator: autolens_workspace + autogalaxy_workspace + autofit_workspace — ALL pre-rewrite remote history
  • confidence: 1.0 (human-authorized leg-7 rewrite, dataset-bulk series)
  • reason: git filter-repo removed dead-at-HEAD blobs (purged datasets, historical output/, old howtolens/ tree, etc.); clones now 32/6/98 MiB (were 107/27/310). Every condemned entry pointing at "remote history" SHAs of these three repos (incl. release-datasets/autolens-regenerable @8625a1de, release-datasets/autogalaxy-regenerable @e940f8cd, and the autofit_workspace SHA 1254a2fe of release-datasets-group-b/all-four — the three HowTo SHAs of that entry survive, HowTo repos were NOT rewritten) is void on the remote.
  • merged: n/a
  • condemned: 2026-07-27
  • sweep-after: never — the backups below are the permanent recover path
  • breaks-if-wrong: nothing further; the rewrite already happened (HEAD trees verified byte-identical, tags preserved by name)
  • archive-ref: permanent local mirror backups ~/Code/PyAutoLabs-backups/{autolens,autogalaxy,autofit}_workspace-pre-rewrite-2026-07-27.git (fsck-verified; old main SHAs 6be18d0cf / f0efa50a9 / 277164bc5)

release-datasets/autocti-imaging-ci

  • type: file
  • locator: autocti_workspace dataset/imaging_ci/{simple,bias_uncorrected,cosmic_rays,non_uniform,parallel_x2__serial_x2,serial_cti} (200 files, 120.2 MB, PR#12)
  • confidence: 0.98
  • reason: committed simulated CTI calibration data, each regenerated in ~10 s by its sole write-site simulator under scripts/imaging_ci/simulators/**; clean-tree regeneration proven per dataset pre-purge and end-to-end post-merge (92 MB rebuilt from empty with a clean git status). Dataset-bulk series leg 6 (issue #11).
  • merged: no
  • condemned: 2026-07-27
  • sweep-after: 2026-08-27
  • breaks-if-wrong: loses the exact historical noise realizations only; regenerated copies are equivalent by construction
  • archive-ref: n/a — committed deletion; pre-purge SHA 272bae3 VOIDED by the 2026-07-27 leg-7 autocti rewrite — bytes recoverable ONLY from the local mirror backup ~/Code/PyAutoLabs-backups/autocti_workspace-pre-rewrite-2026-07-27.git

release-datasets/autocti-dataset-1d-overview

  • type: file
  • locator: autocti_workspace dataset/dataset_1d/{simple,species_x1_continuum_0,species_x3,temporal} + dataset/overview/{dataset_1d,imaging_ci}/ (282 files, 5.35 MB, PR#18)
  • confidence: 0.98
  • reason: committed simulated 1D CTI calibration + overview datasets, each regenerated in seconds by its sole write-site simulator (scripts/dataset_1d/simulators/** and scripts/imaging_ci/simulators/overview/calibrate.py); clean-tree regeneration proven per dataset pre-purge and re-witnessed post-purge with clean git status; 21 auto-simulate guards added to 18 consumers. The dataset-bulk leftover recorded in complete/2026/08/autocti-util-dataset-export.md. The 5 non-simulated overview media files (ccd.gif, ccd_schematic.png, cti.gif, cti_time_evolution.png, what_is_cti.png) stay committed — no write site.
  • merged: no
  • condemned: 2026-08-07
  • sweep-after: 2026-09-07
  • breaks-if-wrong: loses the exact historical noise realizations only; regenerated copies are equivalent by construction
  • archive-ref: n/a — committed deletion; pre-purge SHA 50b701d on autocti_workspace main (post-rewrite history — reachable on the remote)

pyautogalaxy/docs-cite-prodigy

  • type: branch
  • locator: docs/cite-prodigy (PyAutoGalaxy, tip e06a527)
  • confidence: 0.95
  • reason: hygiene tidy pre-scan flagged as a stale local unmerged branch (not the default branch, no unique work in flight) in the PyAutoGalaxy checkout. Human-authorized batch condemnation 2026-07-30.
  • merged: no
  • condemned: 2026-07-30
  • sweep-after: 2026-08-29
  • breaks-if-wrong: loses whatever cite-prodigy docs drafting the branch held; recoverable via pyauto-gut recover pyautogalaxy-docs-cite-prodigy until voided.
  • archive-ref: refs/heads/archive/condemned/pyautogalaxy-docs-cite-prodigy on PyAutoGalaxy origin (e06a527)

pyautolens/docs-cite-prodigy

  • type: branch
  • locator: docs/cite-prodigy (PyAutoLens, tip 03ad3f8)
  • confidence: 0.95
  • reason: hygiene tidy pre-scan flagged as a stale local unmerged branch (not the default branch, no unique work in flight) in the PyAutoLens checkout. Human-authorized batch condemnation 2026-07-30.
  • merged: no
  • condemned: 2026-07-30
  • sweep-after: 2026-08-29
  • breaks-if-wrong: loses whatever cite-prodigy docs drafting the branch held; recoverable via pyauto-gut recover pyautolens-docs-cite-prodigy until voided.
  • archive-ref: refs/heads/archive/condemned/pyautolens-docs-cite-prodigy on PyAutoLens origin (03ad3f8)

pyautobrain/release-accept-red-override

  • type: branch
  • locator: feature/release-accept-red-override (PyAutoBrain, tip 29105d9)
  • confidence: 0.95
  • reason: hygiene tidy pre-scan flagged as a stale local unmerged branch in the PyAutoBrain checkout. Human-authorized batch condemnation 2026-07-30.
  • merged: no
  • condemned: 2026-07-30
  • sweep-after: 2026-08-29
  • breaks-if-wrong: loses the release-accept-red-override prototype work; recoverable via pyauto-gut recover pyautobrain-feature-release-accept-red-override until voided.
  • archive-ref: refs/heads/archive/condemned/pyautobrain-feature-release-accept-red-override on PyAutoBrain origin (29105d9)

pyautobrain/stash-remove-pulse-compat

  • type: stash
  • locator: stash@{0} (PyAutoBrain) — "On remove-pulse-compat: preserve feature/remove-pulse-compat before PyAutoHeart issue 27"
  • confidence: 0.95
  • reason: hygiene tidy pre-scan flagged a stale local stash in the PyAutoBrain checkout. Materialised to a branch (git branch <tmp> stash@{0}, non-destructive — no apply/pop) before archiving, then dropped. Human-authorized batch condemnation 2026-07-30.
  • merged: no
  • condemned: 2026-07-30
  • sweep-after: 2026-08-29
  • breaks-if-wrong: loses the preserved remove-pulse-compat WIP; recoverable via pyauto-gut recover pyautobrain-stash--0- until voided.
  • archive-ref: refs/heads/archive/condemned/pyautobrain-stash--0- on PyAutoBrain origin (ff1af88)

pyautoheart/autonerves-verify-install

  • type: branch
  • locator: feature/autonerves-verify-install (PyAutoHeart, tip 2a1f276)
  • confidence: 0.95
  • reason: hygiene tidy pre-scan flagged as a stale local unmerged branch in the PyAutoHeart checkout. Human-authorized batch condemnation 2026-07-30.
  • merged: no
  • condemned: 2026-07-30
  • sweep-after: 2026-08-29
  • breaks-if-wrong: loses the autonerves-verify-install prototype work; recoverable via pyauto-gut recover pyautoheart-feature-autonerves-verify-install until voided.
  • archive-ref: refs/heads/archive/condemned/pyautoheart-feature-autonerves-verify-install on PyAutoHeart origin (2a1f276)

pyautomind/lifecycle-record-auto-index

  • type: branch
  • locator: feature/lifecycle-record-auto-index (PyAutoMind, tip 4b06f15)
  • confidence: 0.95
  • reason: hygiene tidy pre-scan flagged as a stale local unmerged branch in the PyAutoMind checkout. Human-authorized batch condemnation 2026-07-30.
  • merged: no
  • condemned: 2026-07-30
  • sweep-after: 2026-08-29
  • breaks-if-wrong: loses the lifecycle-record-auto-index prototype work; recoverable via pyauto-gut recover pyautomind-feature-lifecycle-record-auto-index until voided.
  • archive-ref: refs/heads/archive/condemned/pyautomind-feature-lifecycle-record-auto-index on PyAutoMind origin (4b06f15)

pyautomind/morning-status-workflow

  • type: branch
  • locator: feature/morning-status-workflow (PyAutoMind, tip 13ecca7)
  • confidence: 0.95
  • reason: hygiene tidy pre-scan flagged as a stale local unmerged branch in the PyAutoMind checkout. Human-authorized batch condemnation 2026-07-30.
  • merged: no
  • condemned: 2026-07-30
  • sweep-after: 2026-08-29
  • breaks-if-wrong: loses the morning-status-workflow prototype work; recoverable via pyauto-gut recover pyautomind-feature-morning-status-workflow until voided.
  • archive-ref: refs/heads/archive/condemned/pyautomind-feature-morning-status-workflow on PyAutoMind origin (13ecca7)

potential-correction-validation-artifacts

  • type: file
  • locator: ~/Code/PyAutoLabs/potential_correction_validation_artifacts/ (19 files, 688K — loose workspace-root dir, was in no git repo)
  • confidence: n/a — not condemned; archived as a durable BACKUP on explicit human request 2026-08-19
  • reason: artifacts of the closed potential-correction-validation campaign (PyAutoLens#672, closed 2026-08-01): phase 2 JAX-vs-Python parity results (phase2_results/.npz), phase 3 evidence-grid results (subhalo_recovery_evidence_.npz, iter_point_*.npz), diagnostic/run scripts, chain logs, and the phase 4 algorithm_review_report.md. Moved off the workspace root into the Gut attic; local copy deleted after ls-remote verification.
  • merged: n/a
  • condemned: 2026-08-19
  • sweep-after: never — backup, void only on explicit human request
  • breaks-if-wrong: loses the campaign's raw numerical evidence and the algorithm review report; recoverable via pyauto-gut recover potential-correction-validation-artifacts (run from the PyAutoGut checkout), then check out the branch to reabsorb.
  • archive-ref: refs/heads/archive/condemned/potential-correction-validation-artifacts on PyAutoGut origin (716826e, orphan commit, 19 files)

pyautomind/stash-pre-sync-2026-07-06

  • type: stash
  • locator: stash@{0} (PyAutoMind) — "pre-sync local PyAutoMind edits 2026-07-06 (restored after accidental drop)"
  • confidence: 0.95
  • reason: hygiene tidy pre-scan flagged a stale local stash in the PyAutoMind checkout. Materialised to a branch (git branch <tmp> stash@{0}, non-destructive — no apply/pop) before archiving, then dropped. Human-authorized batch condemnation 2026-07-30.
  • merged: no
  • condemned: 2026-07-30
  • sweep-after: 2026-08-29
  • breaks-if-wrong: loses the pre-sync 2026-07-06 WIP; recoverable via pyauto-gut recover pyautomind-stash--0- until voided.
  • archive-ref: refs/heads/archive/condemned/pyautomind-stash--0- on PyAutoMind origin (9dc61c5)

pyautohands/pre-2023-history

  • type: branch
  • locator: master (PyAutoLabs/PyAutoHands, tip 55da101c — 442 commits, 2021-02-12 → 2022-11-27)
  • confidence: n/a — not condemned as trash; archived as durable ORIGINAL PROJECT HISTORY on explicit human request 2026-08-26, then removed from PyAutoHands.
  • reason: the pre-2023 history of PyAutoHands, orphaned by a later history reset. Shares no common ancestor with origin/main (git merge-base exits 1) and is pinned by no tag, so main contains none of its 442 commits and nothing else anywhere reaches them. Authors: Richard 261, Jonathan Frawley 91, James Nightingale 90. The 2026-08-25 org-wide audit flagged it and deliberately left it in place; removed here only once the archive was independently verified.
  • merged: no
  • condemned: 2026-08-26
  • sweep-after: never — original project history, void only on explicit human request
  • breaks-if-wrong: loses the entire 2021-2022 development history of PyAutoHands, including 261 commits authored by Richard Hayes and 91 by Jonathan Frawley. Recoverable via pyauto-gut recover pyautohands-pre-2023-history (run from the PyAutoGut checkout), then check out the branch to reabsorb.
  • archive-ref: refs/heads/archive/condemned/pyautohands-pre-2023-history on PyAutoGut origin (55da101c, 442 commits) — verified by independent clone-back: fetched tip matches, rev-list --count = 442, oldest 7d8743a (2021-02-12), newest 55da101 (2022-11-27). Note: PyAutoBrain branch_archive.yml could NOT do this — PAT_PYAUTOLABS lacks the workflow scope and GitHub rejects any PAT push that introduces .github/workflows/release.yml; archived from a local checkout with a workflow-scoped token instead.

pyautomind/claude-session-branches-2026-08-26-batch

  • type: branch
  • locator: 37 claude/* session branches on PyAutoLabs/PyAutoMind (2026-06-20 → 2026-08-25)
  • confidence: 0.95
  • reason: abandoned Claude Code session branches, each carrying one or a few prompt-filing commits that never landed. Verified NOT silently merged: none had a PR, and for every branch whose touched paths exist on main today, zero files were byte-identical — a later session filed a different prompt at the same path, so the path collides but the content does not. Human-authorized batch condemnation 2026-08-26 ("remove nearly all PyAutoMind ones"). The 3 archive/condemned/* Gut-transit refs and today's live claude/gradient-eager-jit-divergence-py313-5dylgx were excluded.
  • merged: no
  • condemned: 2026-08-26
  • sweep-after: 2026-11-24
  • breaks-if-wrong: loses the unlanded prompt drafts and mind-state edits on those 37 branches. Each is recoverable via pyauto-gut recover <name> until voided; names are pyautomind-<branch-suffix>.
  • archive-ref: 37 refs under refs/heads/archive/condemned/pyautomind-* on PyAutoGut origin — all 37 verified by fresh ls-remote SHA match against the pre-delete tips before any deletion. Full branch→SHA map:
    • claude/pyautoprompt-docs-review-2af7eg @ c7d623618 → archive/condemned/pyautomind-pyautoprompt-docs-review-2af7eg (2026-06-20)
    • claude/health-agent-implementation-x3z8w6 @ 23165376f → archive/condemned/pyautomind-health-agent-implementation-x3z8w6 (2026-06-29)
    • claude/weak-package-review-016kv1 @ 75e918799 → archive/condemned/pyautomind-weak-package-review-016kv1 (2026-07-15)
    • claude/rename-pyautobuild-pyautohands-xqn0p4 @ f80063350 → archive/condemned/pyautomind-rename-pyautobuild-pyautohands-xqn0p4 (2026-07-18)
    • claude/rename-pyautoconf-pyautomerves-i7w9f7 @ be4ca3f53 → archive/condemned/pyautomind-rename-pyautoconf-pyautomerves-i7w9f7 (2026-07-18)
    • claude/autolens-free-ai-setup-jwpyjy @ ae4822372 → archive/condemned/pyautomind-autolens-free-ai-setup-jwpyjy (2026-08-01)
    • claude/autolens-guard-small-datasets-jaqyxv @ d4e4e0dad → archive/condemned/pyautomind-autolens-guard-small-datasets-jaqyxv (2026-08-03)
    • claude/health-agent-full-run-myysos @ 2ebc0f3ae → archive/condemned/pyautomind-health-agent-full-run-myysos (2026-08-05)
    • claude/hygiene-agent-run-n9qtd5 @ 89f540454 → archive/condemned/pyautomind-hygiene-agent-run-n9qtd5 (2026-08-05)
    • claude/interferometer-oom-nufft-yyz98z @ 3e87583b3 → archive/condemned/pyautomind-interferometer-oom-nufft-yyz98z (2026-08-05)
    • claude/markdown-sigma-min-debt-82rpcb @ 64a61f697 → archive/condemned/pyautomind-markdown-sigma-min-debt-82rpcb (2026-08-07)
    • claude/pyautomind-simple-issue-rvfp2b @ f6cf6419c → archive/condemned/pyautomind-pyautomind-simple-issue-rvfp2b (2026-08-07)
    • claude/wake-up-7c3ua4 @ 2f91a2221 → archive/condemned/pyautomind-wake-up-7c3ua4 (2026-08-11)
    • claude/wake-up-z3y6zi @ 2d47239ea → archive/condemned/pyautomind-wake-up-z3y6zi (2026-08-15)
    • claude/clipper-search-identifier-4cowi2 @ 332890828 → archive/condemned/pyautomind-clipper-search-identifier-4cowi2 (2026-08-18)
    • claude/constantzeroth-regularization-cleanup-k1vd0j @ 04731a656 → archive/condemned/pyautomind-constantzeroth-regularization-cleanup-k1vd0j (2026-08-18)
    • claude/multi-galaxy-autolens-package-g1abb3 @ 9a00355c7 → archive/condemned/pyautomind-multi-galaxy-autolens-package-g1abb3 (2026-08-18)
    • claude/wake-up-x5emxb @ ca56cd9af → archive/condemned/pyautomind-wake-up-x5emxb (2026-08-19)
    • claude/autolens-numba-profiling-32su4s @ 9795eddb1 → archive/condemned/pyautomind-autolens-numba-profiling-32su4s (2026-08-21)
    • claude/bilinear-rtu-mesh-docs-pbhsxg @ e70134dfa → archive/condemned/pyautomind-bilinear-rtu-mesh-docs-pbhsxg (2026-08-21)
    • claude/pyautoarray-community-triage-mh7gnp @ 14ce1f82b → archive/condemned/pyautomind-pyautoarray-community-triage-mh7gnp (2026-08-21)
    • claude/autoarray-pixel-scales-int-tuple-wfxlnj @ e69e01a4c → archive/condemned/pyautomind-autoarray-pixel-scales-int-tuple-wfxlnj (2026-08-22)
    • claude/community-release-notification-q4ls6e @ 7f3e6225d → archive/condemned/pyautomind-community-release-notification-q4ls6e (2026-08-22)
    • claude/message-log-partition-tuple-shape-p7uylt @ 466416546 → archive/condemned/pyautomind-message-log-partition-tuple-shape-p7uylt (2026-08-22)
    • claude/autolens-point-jax-vmap-parity-htyfi1 @ a2b051444 → archive/condemned/pyautomind-autolens-point-jax-vmap-parity-htyfi1 (2026-08-23)
    • claude/backport-per-script-timeout-r3w1sv @ 785698b4c → archive/condemned/pyautomind-backport-per-script-timeout-r3w1sv (2026-08-23)
    • claude/slowest-smoke-gate-perf-ovsuay @ ce2217c56 → archive/condemned/pyautomind-slowest-smoke-gate-perf-ovsuay (2026-08-23)
    • claude/stale-jax-pin-smoke-g805rn @ 309d27f1c → archive/condemned/pyautomind-stale-jax-pin-smoke-g805rn (2026-08-23)
    • claude/api-drift-ellipse-grid-lines-8sfjjk @ dba72e390 → archive/condemned/pyautomind-api-drift-ellipse-grid-lines-8sfjjk (2026-08-24)
    • claude/cti-ci-standardisation-phase-6-8uqpfd @ 8abb854dd → archive/condemned/pyautomind-cti-ci-standardisation-phase-6-8uqpfd (2026-08-24)
    • claude/pyauto-cti-ci-phase-5-n4idom @ 332c4513d → archive/condemned/pyautomind-pyauto-cti-ci-phase-5-n4idom (2026-08-24)
    • claude/pyautohands-rejects-autocti-bug-vu5ulq @ 5c0ae8bea → archive/condemned/pyautomind-pyautohands-rejects-autocti-bug-vu5ulq (2026-08-24)
    • claude/pynufft-removal-phase-3-ird7b8 @ 15abb314f → archive/condemned/pyautomind-pynufft-removal-phase-3-ird7b8 (2026-08-24)
    • claude/strong-lensing-papers-2026-6cpv0a @ f4865cebc → archive/condemned/pyautomind-strong-lensing-papers-2026-6cpv0a (2026-08-24)
    • claude/test-mode-bypass-assertion-ties-jqi2f1 @ 96143497a → archive/condemned/pyautomind-test-mode-bypass-assertion-ties-jqi2f1 (2026-08-24)
    • claude/loggaussian-prior-support-buv5xe @ 5dfffb4c3 → archive/condemned/pyautomind-loggaussian-prior-support-buv5xe (2026-08-25)
    • claude/pyautohands-rejects-autocti-bug-mk1yi9 @ 85f6df2af → archive/condemned/pyautomind-pyautohands-rejects-autocti-bug-mk1yi9 (2026-08-25)

pyautohands/2022-era-feature-branches

  • type: branch
  • locator: feature/dynamic_scripts (tip 3f5eaf0, 313 commits) and feature/build_tests (tip 0c1ad5d, 312 commits) on PyAutoLabs/PyAutoHands
  • confidence: n/a — not condemned as trash; removed alongside [[pyautohands/pre-2023-history]] on explicit human request 2026-08-26.
  • reason: siblings of the archived pre-2023 master — both share its ancestry and are likewise disjoint from today's main. feature/build_tests carried 0 commits not already reachable from the master archive, so it needed no archive of its own. feature/dynamic_scripts carried 9 unique commits (all Richard, 2022-10-10, CI iteration: "install and use jq", "checkout", "cd", "remove commented step") that the master archive did NOT cover, so it was archived separately before deletion.
  • merged: no
  • condemned: 2026-08-26
  • sweep-after: never — 2022-era project history, void only on explicit human request
  • breaks-if-wrong: build_tests loses nothing (fully contained in the master archive). dynamic_scripts would lose its 9 unique CI commits; recoverable via pyauto-gut recover pyautohands-feature-dynamic-scripts.
  • archive-ref: refs/heads/archive/condemned/pyautohands-feature-dynamic-scripts on PyAutoGut origin (3f5eaf0, 313 commits) — verified by independent clone-back before the source branches were deleted. feature/build_tests: n/a, its content is reachable from archive/condemned/pyautohands-pre-2023-history (verified git rev-list --count 55da101c..build_tests = 0); pre-delete SHA 0c1ad5d.

autolens_profiling/jax-optimizer-tuning

  • type: branch
  • locator: feature/jax-optimizer-tuning (tip fc270a2, 2 commits ahead, last touched 2026-07-15) on PyAutoLabs/autolens_profiling
  • confidence: 0.95
  • reason: the harness work behind issue #69 ("tune the JAX multi-start optimizers into a standard option"), which was CLOSED as superseded on 2026-08-18 (human-confirmed) by the 2026-08-17 inference programme (#134, results/notes/inference/PROGRAMME.md, merged via #136/#137). The learning_rate axis dissolved into the lr-free MultiStartProdigy; the Sersic lens + Sersic source case was deliberately dropped by human decision. Never opened a PR, and the searches/ tree it edits no longer exists on main at all. Mind record: complete/archive/shelved/jax_optimizer_settings_tuning.md.
  • merged: no
  • condemned: 2026-08-26
  • sweep-after: 2026-11-26
  • breaks-if-wrong: loses 181 lines of settings-grid/truth-scoring harness across searches/_samplers.py, searches/_setup.py, searches/sweep.py and two imaging/sersic.py cells. Recoverable via pyauto-gut recover autolens-profiling-jax-optimizer-tuning. Note the layout mismatch: searches/ is gone from main, so a reabsorb is a port, not a merge.
  • archive-ref: refs/heads/archive/condemned/autolens-profiling-jax-optimizer-tuning @ fc270a2ffbf73393f6c1d85b2cb90d7a0e551bc6 on PyAutoGut — verified by independent clone-back (102 commits) before the source branch was deleted.

autolens_profiling/multiband-compile-ab

  • type: branch
  • locator: archive/condemned/multiband-compile-ab (tip 7735151, 3 commits ahead, last touched 2026-07-21) on PyAutoLabs/autolens_profiling
  • confidence: 0.9
  • reason: a jax_compile multi-band A/B experiment (FactorGraphModel vag compile; the finding that lax.map scan is the compile killer and pyloop fixes it) that someone had already named into the archive namespace in the wrong repo — the archive lives on PyAutoGut, not on the source repo — and never filed a condemned.md entry, so nothing pinned it. This entry is that missing record; the payload has been re-homed to the Gut where the namespace actually means something.
  • merged: no
  • condemned: 2026-08-26
  • sweep-after: never — its README carries the lax.map/pyloop compile finding, which is a result rather than spent scaffolding; void only on explicit human request
  • breaks-if-wrong: loses 319 lines — jax_compile/README.md (113 lines of experiment write-up), jax_compile/results/local_cpu/mge.json measurements, and instruments/imaging.py / searches/_setup.py harness. Recoverable via pyauto-gut recover autolens-profiling-multiband-compile-ab.
  • archive-ref: refs/heads/archive/condemned/autolens-profiling-multiband-compile-ab @ 7735151b89147c09d29ec98480fb03fb2bb3d21b on PyAutoGut — verified by independent clone-back (130 commits) before the source branch was deleted.

autolens_profiling/stash-strip-non-jit-noise

  • type: stash
  • locator: stash@{0} in the autolens_profiling canonical checkout — "On strip-non-jit-noise-add-alma-high-res: strip-noise wip", dated 2026-06-07
  • confidence: 0.95
  • reason: 11 weeks untouched; its base branch strip-non-jit-noise-add-alma-high-res no longer exists on origin, and the likelihood/ scripts it edits are not on main's current tree. The diff is a strip/simplify pass (325 insertions, 778 deletions across 9 likelihood scripts — removing per-channel eager numpy baselines in favour of the JIT loop), not unlanded new work.
  • merged: no
  • condemned: 2026-08-26
  • sweep-after: 2026-11-26
  • breaks-if-wrong: loses the noise-stripping edits to likelihood/{datacube,imaging,interferometer,point_source}/*.py. Materialised as a real commit before the drop, per the schema's rule that a manifest merely pointing at a stash is worthless once it is dropped. Recoverable via pyauto-gut recover autolens-profiling-stash-strip-non-jit-noise.
  • archive-ref: refs/heads/archive/condemned/autolens-profiling-stash-strip-non-jit-noise @ e6e72cc2dfd67a7ae5cb2fff3b0fb0ae3dc56e56 on PyAutoGut — verified by independent clone-back (9 files, 325/778 diff intact) before git stash drop.

autolens_workspace_developer/rectangular-spline-rotated-experiments

  • type: file
  • locator: rect_adapt_duo/ (7 files, incl. bundled dataset/*.fits), searches_minimal/probe_grad_pix_adapt_image.py, jax_profiling/misc/pixelization_spline_vs_linear.py, jax_profiling/misc/pixelization_spline_fit_comparison.py, jax_profiling/results/jit/imaging/spline_vs_linear_fit/ (2 PNGs), jax_profiling/results/jit/imaging/spline_vs_linear_hst_v2026.4.13.6.json, ..._sweep.png — 14 files, 1775 lines plus binaries, in PyAutoLabs/autolens_workspace_developer
  • confidence: 0.97
  • reason: every one of these exercises a mesh class that no longer exists in PyAutoArray. The 2026-07-23 rectangular-mesh consolidation (PyAutoArray#402/#403) deleted RectangularSplineAdapt{Density,Image} and RectangularRotatedAdaptImage; these scripts import them at module scope and cannot run at all against any main since. They are spent comparison experiments — the spline and rotated transforms lost to the consolidation, which chose the kernel-CDF transform instead. Not wrong code, just code whose subject the library removed. Task: autolens_workspace_developer#131, prompt active/rectangular_experiments_gut_stash.md.
  • merged: no
  • condemned: 2026-08-27
  • sweep-after: n/a — nothing to void. These are committed deletions on a pushed repo, so the bytes stay reachable in remote history permanently; per the entry schema's "Committed code / test deletions" rule this record exists only to pin the pre-delete SHA. There is no ref to expire and no pyauto-gut void step.
  • breaks-if-wrong: loses the rotated-vs-spline mesh demo (rect_adapt_duo/compare_meshes.py + its self-contained simulated dataset), the spline adapt-image FD gradient probe, and the spline-vs-linear pixelization profiling comparison with its measured HST results (JSON + sweep plot). The findings those runs produced survive in jax_profiling/gradient/README.md and the PyAutoArray#402 completion record; what is lost is the ability to re-run them, which already requires resurrecting three deleted library classes.
  • archive-ref: n/a — committed deletion; pre-delete SHA 9ae0502 on PyAutoLabs/autolens_workspace_developer main (the base of feature/rectangular-experiments-gut-stash). Recover with git checkout 9ae0502 -- <path>.