Build structured hierarchy rows and folded paths before rendering ASCII or Mermaid. Move compact and detailed diff writers alongside their text presentation helpers.
Use signed cell rows and keep whole-cell contents out of the text diff. Group reference changes, fold redundant terminal edges into the cell list, and inline one-off layer names. Count identical unmatched objects while retaining occurrence-based display limits.
Summarize added and removed cells, group references, and keep replacement pairs together at text display limits. Trim hierarchy labels and summarize long unmatched paths.
Stream JSON object evidence and reduce parsing and comparison allocations. Parse large revisions concurrently and adaptively reuse matching cell bodies when most of the layout is shared.
Share record framing, name decoding, payload validation, array reference
decoding, and Arrow cell-header parsing. Accept trailing NUL padding in
names and retain located diagnostics for malformed records.
Summarize text diffs by cell, layer, and reference target. Show comparison
and omission counts before detail. Compact canonical dumps while
preserving exact fields and duplicate counts for Git textconv. Simplify
hierarchy output and retain per-object evidence in JSON.
Detect gzip headers on every CLI input and decode all members before parsing.
Keep report offsets relative to the decompressed stream and report corrupt
archives as invalid input. Recognize OASIS headers and explain how to provide
GDSII data.
Cover mixed compressed and plain comparisons, corrupt trailers, and layer-name
output from compressed input.
Capture revision metadata, cell presence and attributes, and structured
diagnostics in owned comparison results. Derive outcomes and render text,
JSON, and hierarchy reports without retaining source documents. Keep derived
descriptions and graph analysis in their renderers.
Use pheon-structural-v1 and PHEON_LAYER_NAMES_V1 consistently, expose helpers
through their defining modules, and remove obsolete re-exports. Update user
guides and report fixtures without changing schema version 1.
Cover dropped inputs, diagnostic ordering, absent and empty cells, offset
independence, and CLI failures. All 92 Rust tests pass in debug and release;
report and hierarchy goldens retain their behavior.
Implement native GDSII reading and writing with Arrow interchange, hierarchy
scans and selective cell loading. Validate numeric packing and array pitches,
and preserve normalized REAL8 encodings exactly within the supported range.
Add standalone dump, structural diff, hierarchy, layer-name annotation and Git
diff commands. Provide scoped text and JSON reports, deterministic comparison,
and record-preserving metadata edits.
Package the shared library and optional Python Arrow readers as pheon using
CFFI ABI mode and Python-independent platform wheels. Include locked source
builds, native API checks, cross-platform artifact workflows, documentation
and regression coverage. Require Rust 1.88 or newer and Python 3.11 or newer.
Validation: 87 native tests, 14 binding tests on Python 3.11/3.13/3.14,
and 71 installed companion integration tests pass. Full, scanned and selected
Arrow layouts match the previous schemas.