Csv

Datagrunt 5.3.10: Wave 2B Pooled HTTP Keep-Alive, Single-Pass Cache, Min-Heap Runner & Streaming Batch

Datagrunt 5.3.10 delivers Wave 2B performance optimizations across the rationalization subsystem—featuring pooled HTTP/1.1 keep-alive connection reuse, single-pass fast request key generation, min-heap rate-key scheduling, streaming batch JSONL I/O, and shared envelope isolation.

Read more →

November 13, 2026

Datagrunt 5.3.9: Wave 2A Consensus, Grounding, and Folding Optimizations + 5 Bug Fixes

Datagrunt 5.3.9 delivers Wave 2A performance optimizations across consensus, grounding, and folding subsystems—featuring inverted-index unkeyed record matching, copy-on-write null stripping, direct leaf wire serialization—along with 5 key bug fixes.

Read more →

November 13, 2026

Datagrunt 5.3.8: Rationalization Hot-Path Performance and Include Schema Cache Collision Fix

Datagrunt 5.3.8 introduces major hot-path performance optimizations across schema, unit, envelope, and wire processing, adds build_envelopes_isolated for fault-isolated batch envelope creation, and resolves an include_schema cache key collision bug.

Read more →

November 12, 2026

Datagrunt 5.3.7: Corroborated CAD 3 Tr SHX Drift and Hardened Numeric Guards

Datagrunt 5.3.7 introduces multi-word corroborated CAD 3 Tr SHX kerning drift resolution, 2-letter preposition bridging, EU directive year detection, and hardened sign/numeric/negation guards.

Read more →

November 11, 2026

Datagrunt 5.3.6: Single-Pass Object Walk, Fast Trusted Constructors, and Single-Load Corpus Renders

Datagrunt 5.3.6 delivers single-pass PDFium object tree walks, fast trusted constructors for path segments, single-load fused page renders, and bounded in-memory prerender caching.

Read more →

November 11, 2026

Datagrunt 5.3.5: Overhang-Verified CAD Edge-Letter Drift and Sparse/MRC Scan Deduplication

Datagrunt 5.3.5 introduces overhang-verified vector stroke recovery for CAD edge-letter SHX drift, alongside sparse tile propagation and multi-layer MRC scan image deduplication.

Read more →

November 10, 2026

Datagrunt 5.3.4: Word-Level CAD Tr 3 Backing, Searchable Scan Tile Deduplication, and Graham LPT Corpus Scheduling

Datagrunt 5.3.4 delivers word-level vector stroke backing for CAD Text Render Mode 3 (Tr 3) text callouts, advanced tiled raster figure deduplication for searchable scans, Graham LPT scheduling for parallel corpus processing, and parallel pytest-xdist CI with PEP 669 sysmon coverage.

Read more →

November 10, 2026

Datagrunt 5.3.3: Multi-Core Corpus Processing, CAD Vector Text, and Fast 16MP Renders

Datagrunt 5.3.3 introduces multi-core process-isolated corpus extraction with smart bundle reuse, vector-backed CAD Text Render Mode 3 (Tr 3) text and table extraction, mosaic/tiled figure deduplication, zero-copy 16MP PNG renders, and PDFium session batching.

Read more →

November 10, 2026

Datagrunt 5.3.2: PDF Stream Line Assembly & CI Workflow Deduplication

Datagrunt 5.3.2 introduces bounded stream layout line assembly (_OpenLine) to prevent tall glyphs from merging adjacent lines, serializes version bump releases (#483), deduplicates post-merge push CI runs, and updates action runners.

Read more →

November 9, 2026

Datagrunt 5.3.1: In-Process Fast Path, Multimodal VLM Wire & Non-Destructive Consensus

Datagrunt 5.3.1 introduces in-process zero-loss fast paths for PDF compute, 3-witness multimodal VLM band wire (#473), out-of-prompt OCR grounding, non-destructive consensus and folding (#482), Informational markers, and conditional 4-way orientation voting.

Read more →

November 9, 2026

Tags