fix(render): refresh trace render when a trace's attributes change#69
Merged
Conversation
The incremental field render (TraceLayer.generateTraceLayer with window_moved=False, used when generate_image=False) redraws from a cached traces_in_view list of trace OBJECTS, reconciled only by section.added_traces / removed_traces. Section.editTraceAttributes - the single funnel for color/name/tags/fill edits from the trace dialog, trace list, object list, paste-attributes and bulk/import recolors - REPLACES the trace with a fresh copy rather than mutating in place. A table refresh calls section.clearTracking(), emptying the add/remove lists, after which an incremental render still holds the old, replaced object and draws its stale color. The selection highlight, drawn from the live selected_traces set, tracked the new color - producing the reported mismatch (highlight new, line old). Bug is inherited from upstream, not a fork regression. Guard the incremental branch: if any cached on-screen trace is no longer a live member of its contour, fall back to a full tracesAsList() rebuild so current attributes are always drawn. Only the on-screen subset is checked, so a selection-only refresh stays on the fast path (no full rebuild), preserving the large-series render optimization. Adds tests/test_trace_color_stale_render.py covering two color-change paths (editTraceAttributes and object-list-style bulk recolor), the exact line/highlight mismatch, a sibling name-change, and a perf guard asserting selection-only refreshes never rebuild. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
dustenhubbard
enabled auto-merge (squash)
July 18, 2026 02:06
dustenhubbard
added a commit
that referenced
this pull request
Jul 18, 2026
Covers the trace-color render fix (#69), Windows undo UTF-8 fix and checkbox/CI hardening (#70), upstream interop escaping (#71), and the zarr conversion CPU bound (#72). Moves the zarr note from the beta-3 section (already shipped) into the new beta-4 section. Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
dustenhubbard
added a commit
that referenced
this pull request
Jul 19, 2026
Objects imported before the autoseg color features baked their colors in at import time, so they never pick up the current colorblind-safe default (or a custom) palette. Add "Reapply autoseg colors..." to the object-list / object context menu to push the CURRENT palette + seed back onto selected objects. - palette.py: AUTOSEG_TRACE_PREFIX (single source of truth for the "autoseg_<id>" import naming), label_id_from_name (recovers the label id from an unmodified import name), and palette_color_for_name (reproduces the exact import color for a parseable name; falls back to a stable, PYTHONHASHSEED- independent zlib.crc32 hash of the name otherwise). conversions.py now builds import names from the shared prefix so the two stay in sync. - series.reapplyAutosegColors: resolves the palette/seed and per-object color once, then rewrites colors through the normal per-section editTraceAttributes path over only the sections the objects appear on -- one undoable operation, correct list/field refresh (reuses the #69 stale-render funnel), no per-object full-series scans. - GUI action goes through the existing object_function wrapper, so locked objects are blocked exactly as any other bulk attribute edit, and a confirm dialog warns before existing colors are discarded (undo restores them). Selection is the contract: it composes with the regex/group/invert filters, so "select all autoseg_* then reapply" works without guessing a name pattern. Tests: id-parse + hash-fallback determinism, custom palette/seed respected, recolor matches a fresh import, distinct ids stay distinct, single undo restores every prior color across sections, zero-selection no-op. Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
dustenhubbard
added a commit
that referenced
this pull request
Jul 23, 2026
The incremental field render (TraceLayer.generateTraceLayer with window_moved=False, used when generate_image=False) redraws from a cached traces_in_view list of trace OBJECTS, reconciled only by section.added_traces / removed_traces. Section.editTraceAttributes - the single funnel for color/name/tags/fill edits from the trace dialog, trace list, object list, paste-attributes and bulk/import recolors - REPLACES the trace with a fresh copy rather than mutating in place. A table refresh calls section.clearTracking(), emptying the add/remove lists, after which an incremental render still holds the old, replaced object and draws its stale color. The selection highlight, drawn from the live selected_traces set, tracked the new color - producing the reported mismatch (highlight new, line old). Bug is inherited from upstream, not a fork regression. Guard the incremental branch: if any cached on-screen trace is no longer a live member of its contour, fall back to a full tracesAsList() rebuild so current attributes are always drawn. Only the on-screen subset is checked, so a selection-only refresh stays on the fast path (no full rebuild), preserving the large-series render optimization. Adds tests/test_trace_color_stale_render.py covering two color-change paths (editTraceAttributes and object-list-style bulk recolor), the exact line/highlight mismatch, a sibling name-change, and a perf guard asserting selection-only refreshes never rebuild.
dustenhubbard
added a commit
that referenced
this pull request
Jul 23, 2026
Objects imported before the autoseg color features baked their colors in at import time, so they never pick up the current colorblind-safe default (or a custom) palette. Add "Reapply autoseg colors..." to the object-list / object context menu to push the CURRENT palette + seed back onto selected objects. - palette.py: AUTOSEG_TRACE_PREFIX (single source of truth for the "autoseg_<id>" import naming), label_id_from_name (recovers the label id from an unmodified import name), and palette_color_for_name (reproduces the exact import color for a parseable name; falls back to a stable, PYTHONHASHSEED- independent zlib.crc32 hash of the name otherwise). conversions.py now builds import names from the shared prefix so the two stay in sync. - series.reapplyAutosegColors: resolves the palette/seed and per-object color once, then rewrites colors through the normal per-section editTraceAttributes path over only the sections the objects appear on -- one undoable operation, correct list/field refresh (reuses the #69 stale-render funnel), no per-object full-series scans. - GUI action goes through the existing object_function wrapper, so locked objects are blocked exactly as any other bulk attribute edit, and a confirm dialog warns before existing colors are discarded (undo restores them). Selection is the contract: it composes with the regex/group/invert filters, so "select all autoseg_* then reapply" works without guessing a name pattern. Tests: id-parse + hash-fallback determinism, custom palette/seed respected, recolor matches a fresh import, distinct ids stay distinct, single undo restores every prior color across sections, zero-selection no-op.
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.
Fixes the beta-reported bug where changing a trace's color updated the selection highlight but left the drawn line in the old color.
Root cause (inherited from upstream, 2023): the incremental field render reuses cached trace objects reconciled only via
added_traces/removed_traces;editTraceAttributescopy-replaces the trace, and a table-refreshclearTracking()in between leaves the cache holding the stale object. Highlight draws from the live selection set, hence the mismatch.Fix: self-healing validation in the incremental branch — cached on-screen traces are checked against ground truth (identity in their contour); any staleness triggers a rebuild. Covers all copy-replace attribute edits (color, name, tags, fill). Selection-only refreshes stay on the fast path (perf-asserted: zero rebuilds).
Deterministic headless pixel repro: 4 new tests fail unpatched, pass patched; full suite green.
🤖 Generated with Claude Code