Skip to content

TechEngine roadmap and status #97

Open

Description

@Seungpyo1007

Purpose

Roadmap and status board for TechEngine, the engine behind the TechAPI dataset: validation, verification, scoring, the static dump, collectors, and TechEngineBot's checks on TechAPI PRs.

This issue stays open. The body is the current status. Comments are one summary per milestone or major change. Day-to-day work is tracked in the workstream issues below and on Project #2 (Workstream field).

How work is tracked (since 2026-09-27)

Workstream Issue Covers
Roadmap this issue Direction, milestones, status
Verification & TechEngineBot #98 app/verify, scoring, integrity gate, bot PR comments, promotion
Ingest & sources #99 app/ingest collectors and parsers, enrich benchmark sources, weekly-ingest
Dump, API & CI #100 app/dump, schemas/serializers, validate.py, workflows, submodule sync

TechEngine has no automatic tracker routing. PRs link their workstream with Refs #N. main is the default branch here, so Closes #N in a commit or body really closes the tracker (#101 closed #98 once). Work that concerns TechAPI data also references GetTechAPI/TechAPI#297. A bare #1 means TechEngine #1.

Current status (2026-09-27)

  • Coverage: since feat: analyze all dataset categories and actual PR diffs #95, every bot check covers 12/12 categories (191,797 records) and diffs against the PR's real base. Before that it was 8 categories, and the integrity gate checked only CPU/GPU (3.4%).
  • Tests: 445 passing (pytest), plus mypy and ruff clean. CI test runs on every PR.
  • Verification (TechAPI): 23,403 records verified: 21.4% of device categories (109,237), 12.2% of all 12 categories (191,814). Laptop and monitor have domain rules since feat(verify): add laptop and monitor domain scoring #102 (103 laptops promoted); software and website have structural checks only. The 996 red SoCs have no failing check; they are placeholder imports with weak sources, and red is intended.
  • Dump: deterministic (slug-hash ids, git-history timestamps). About 33 minutes locally for 191k records. Phone ram_gb accepts sub-GB values since Store phone RAM as a float so sub-GB phones can be served #94.
  • Branch policy: trunk-based main. No branch protection: TechEngineBot and the maintainer are both admins, so the rule is to merge only after test passes.
  • Scheduled jobs: weekly-refresh had 5 failed or cancelled runs before fix(ci): open weekly refresh PRs against TechAPI develop #96; re-check after the 2026-09-28 run. weekly-ingest, verify-network (Mon), coverage-report, verify-status and bump-techapi are active.

Recent changes

PR Date Change
#102 2026-09-27 Laptop and monitor domain scoring (107 laptops green)
#101 2026-09-27 Categories without domain rules scored over 65 points (no false red)
#96 2026-09-26 Weekly refresh PRs target TechAPI develop
#95 2026-09-26 All categories analyzed; real PR diff base; dump check; scope shown in comments
#94 2026-09-25 Phone RAM stored as float (sub-GB phones)
#93 2026-09-25 Wikipedia GPU multi-row headers; threads no longer guessed from core counts
#92 2026-09-25 Validator accepts sub-GB phone RAM
#90 2026-09-25 Wikipedia CPU parser fixes (fractional TDP, clock ranges, L2, family rows)

Goals (milestone "Engine: full coverage and scoring (2026)")

  1. Domain scoring for laptop and monitor (done, feat(verify): add laptop and monitor domain scoring #102); next software and website.
  2. Laptop/monitor Tier 0 red (fixed, fix(verify): normalize laptop and monitor structural scores #101).
  3. All scheduled workflows green for 4 consecutive weeks.
  4. Fix or explain the 18 CPU anomalies from the full integrity scan.

Metadata

Metadata

Labels

enhancementNew feature or requesttrackingLong-running tracker issue (stays open)

Type

No type

Projects

Relationships

None yet

Development

No branches or pull requests

Issue actions