Skip to content

Preserve site-free aggregate source scope through publication (addresses #728) - #732

Merged
trissim merged 14 commits into
mainfrom
fix/site-free-aggregate-publication-728-20261005
Oct 5, 2026
Merged

trissim merged 14 commits into
mainfrom
fix/site-free-aggregate-publication-728-20261005

Conversation

@trissim

@trissim trissim commented Oct 5, 2026 •

Copy link
Copy Markdown
Collaborator

Fixes #728's acquired-address collision and saved SITE-free aggregate publication/reload. Exact qualified head 2f4c68ff05876efb48e14743ac8027064b1b09ab is now installed/public accepted on a distinct original receiving13 case. This is synthetic engineering acceptance, not biological accuracy or every checkpoint-metadata contract.

Existing owners and deleted competing decisions

  • RuntimeArtifactMaterialization no longer addresses a record by shared memory location. Artifact-named output consumes its full original runtime scope; explicitly source-named writers retain original source identity.
  • PlateImageInventory no longer overwrites admitted source metadata with filename coordinates. SITE-free source metadata survives saved inventory and streaming; ordinary unbound files still use their original filename parser.
  • No duplicate publisher, identity store, codec, fixture-name workaround or guard relaxation. The source repair deletes six net production lines.

Normal integration includes main1715215f8 and merged743/744/753/759. The two repaired source owners are unchanged from the106-control checkpoint. No unmerged Root ancestry or private runtime stack is delivered through this PR.

Source evidence

106 affected source controls pass:79 materialization/address-family and27 inventory/reload, both SITE insertion orders, automatic/default and explicitly source-named writers, saved values, real TIFF loading, SITE absence, spacing and contributors. Focused before/after AST covers14 related owners/consumers,28 ASTs, no parse omissions; no global R1 claim. Original receiving12 failures remain intact.

Actual ordinary installed public path

Receiving13 original compilejob1 c141b006-4713-44ff-846f-357f8b7b0b0f, executejob2 4ffd2a7a-c02c-4590-b906-bcd2dfd4ed95: COMPLETE/errors[], allfour steps succeed. Original read-only installed checker3135 terminated0; actual exported typed runtime/artifact addresses and saved CSVs reconcile:

  • Four acquired SITE1/3 × CHANNEL1/2 records preserve distinct CSV paths/full scopes and exact4×6 raw pixels.
  • Default and explicit aggregate images preserve each channel's4×12/48pixels, address-bound table rows, true SITE1/3 contributors, Y.25/X.5µm calibration and absence of scalar SITE.
  • The SAME owned viewer's ordinary saved image stream admits both channel images/errors[]. Actual native samplers prove all48pixels per channel11/13 and21/23, uint16 and SITE-free metadata/calibration.
  • First saved screenshot retained as unsuitable because an older acquired route remained selected. Supported isolation selected selected_images CHANNEL2; Dewey and Planck personally opened the positive native screenshot with both halves and no saved scalar SITE. Older mounted acquired routes retain their genuine SITE axis independently.

Explicit limit: acquired checkpoint source_voxel_spacing is empty in this case. Aggregate calibration passes, but we do not claim every acquired checkpoint passed spacing. That separate pass-through gap is not hidden or repaired by this PR.

Original sole client40642 EOF terminal0. Exact native3293555/create1791205633.28 and viewer3510427/create1791207058.28 public close acknowledgments report all required closure fields true/errors[], independent PIDs/endpoints absent. Original MCP invocation inactive/dead. Post-writer complete journals are sealed; no execution or UNKNOWN replay, frozen output edits or substitute client.

Actual receipts under engineering-pre-first-routing-20261004/receiving13/public73295:

  • QUALIFICATION-732.rst
  • ORIGINAL-CHECKER-3135.rst (original checker strength; later saved receiving completion is in qualification)
  • OWNER-TERMINAL.rst
  • ADMIN732_REPAIRED13_95/author-workspace/output/screenshots/saved-reopen02-selected/20261005T134227164557Z_napari_6015_OpenHCS_Napari_Visualization.png, SHA256 4ffa6c368f3dd6df960428ddd3bcfbe93bc1f69d41dcf54c08c43cce19834c9e.

Original failed receiving12 case remains COMPLETE execution but failed publication/reload, with its original logs/closures untouched. Current scientific prefixes/skill bundles are unchanged. #748's automatic fractional-Z point/volume-label case and Singer522 selected-source navigation remain separate. Hosted CI is deferred by the user; local/source and real affected installed path are demonstrated.

@trissim

trissim commented Oct 5, 2026

Copy link
Copy Markdown
Collaborator Author

Retained assisted saved-label triage is published at c2dbfdf in the existing receiving receipt. Exact native6013.log43–58 names FIELD site5 labels, not the stitched mosaic. Original saved metadata declares SOURCE_BINDING with two DAPI/FITC source planes, empty contributors and no scalar CHANNEL; manual reopening preserves that plane domain and scalar-layer route construction refuses it. This does not establish that732 fixes saved-label reopening:732 scopes genuine contributor reduction at the publication adapter; the manual path bypasses it.407 is presentation-only. Existing152 producer-family ownership is coordinated directly with Planck/Dewey; Singer retains receiving integration. No competing patch, live replay, science mutation or guard weakening. The tiny saved-label control is a separately pinned producer acceptance alongside the shared receiving packet, not a new732 merge hold. Production/source41 qualification unchanged;732 ordinary installed default-publication acceptance remains pending.

@trissim
trissim marked this pull request as ready for review October 5, 2026 13:53
Copilot AI balanced review requested due to automatic review settings October 5, 2026 13:53
@trissim
trissim merged commit 121fad0 into main Oct 5, 2026
1 of 3 checks passed

Copilot AI left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Site-free acquired mosaic fails automatic viewer publication and compiled table-schema reconciliation

2 participants