azaion/satellite-provider
1
0
Fork 0
mirror of https://github.com/azaion/satellite-provider.git synced 2026-06-21 19:51:14 +00:00
Code Issues Packages Projects Releases Wiki Activity
Files
ac40a8b35240e8ae560f5208c198c54b16fcfba0
satellite-provider/_docs/02_document/architecture_compliance_baseline.md
T
Oleksandr Bezdieniezhnykh 08451df027
ci/woodpecker/push/01-test Pipeline was successful
Details
ci/woodpecker/push/02-build-push Pipeline was successful
Details
[AZ-350] Close 03-code-quality-refactoring: Phase 6+7 + FINAL_report 
Phase 6 (Verification): smoke run green (format gate + 200/200
unit + integration smoke). verification_report.md captures
metric deltas vs Phase 0 baseline; all 5 ACs met, all 4
constraints honored, 0 regressions.

Phase 7 (Documentation):
- module-layout.md: corrected DataAccess->Common dependency
  (was mistakenly documented as "Imports from: (none)" by
  prior AZ-315 baseline; csproj reference + 7 import sites
  have actually been there since AZ-309).
- architecture_compliance_baseline.md: F5 entry revised to
  reflect the actual layering invariant (one-way: Common
  MUST NOT import from DataAccess, but DataAccess MAY
  import from Common).
- 00_discovery.md: added "Updates Since Baseline" section
  enumerating the AZ-309 split + AZ-350 27-change run +
  AZ-372 tooling additions; original tree kept as a
  2026-05-10 snapshot.

FINAL_report: complete run summary (10 batches, 27 tasks,
3 K=3 cumulative reviews, baseline->final metric table,
remaining items, lessons learned).

Autodev state: advance Step 8 -> Step 9 (New Task);
sub_step reset to phase 0 awaiting-invocation.

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-05-11 05:23:35 +03:00

7.4 KiB
Raw Blame History

Architecture Compliance Baseline

Date: 2026-05-10 Mode: Baseline (Phase 1 + Phase 7) Scope: Full existing codebase Verdict: PASS_WITH_WARNINGS (baseline) → all findings resolved by epic AZ-309

Findings

# Severity Category File:Line Title Status
1 High Architecture SatelliteProvider.Services/TileService.cs:11 Concrete dependency on GoogleMapsDownloaderV2 bypasses ISatelliteDownloader Resolved (pre-AZ-309 cleanup)
2 High Architecture SatelliteProvider.Common/Interfaces/ISatelliteDownloader.cs ISatelliteDownloader interface is dead code Resolved (pre-AZ-309 cleanup)
3 Medium Architecture SatelliteProvider.Api/Program.cs:141 API endpoint directly injects concrete downloader + repository Resolved by AZ-310 + AZ-311 (commit 8b0ddae's parent)
4 Medium Architecture SatelliteProvider.Services/ No physical boundary between logical components in shared project Resolved by AZ-312 + AZ-313 + AZ-314 (commit 8b0ddae)
5 Low Architecture module-layout.md DataAccess documented as importing Common but actually has zero cross-project dependencies Resolved by AZ-315 (this commit) — module-layout.md now reflects the actual no-import layout

Finding Details

F1: Concrete dependency on GoogleMapsDownloaderV2 (High / Architecture)

  • Location: SatelliteProvider.Services/TileService.cs:11
  • Description: TileService depends on the concrete class GoogleMapsDownloaderV2 instead of ISatelliteDownloader. DI registration is also concrete (AddSingleton<GoogleMapsDownloaderV2>()). This couples the entire tile pipeline to a single provider.
  • Impact: Adding a new satellite imagery provider requires modifying TileService and Program.cs DI wiring rather than just registering a new implementation.
  • Suggestion: Have GoogleMapsDownloaderV2 implement ISatelliteDownloader, update DI to register via interface, inject interface into TileService.

F2: ISatelliteDownloader is dead code (High / Architecture)

  • Location: SatelliteProvider.Common/Interfaces/ISatelliteDownloader.cs
  • Description: The interface exists (declares GetTiles(GeoPoint, double, int, CancellationToken)) but NO class implements it and NO code references it. The actual downloader method used is GoogleMapsDownloaderV2.GetTilesWithMetadataAsync() which has a different signature.
  • Impact: The provider-agnostic abstraction doesn't function. Interface and implementation have diverged.
  • Suggestion: Update ISatelliteDownloader to match the actual API surface needed by consumers, then implement it in GoogleMapsDownloaderV2.

F3: API endpoint bypasses service layer (Medium / Architecture) — RESOLVED

  • Location: SatelliteProvider.Api/Program.cs:141 (ServeTile) and :206 (GetTileByLatLon)
  • Description: Two API endpoints directly inject GoogleMapsDownloaderV2 and ITileRepository instead of using ITileService. This bypasses the service layer and creates a shortcut from Layer 4 to Layer 2.
  • Impact: Business logic (caching, dedup) in TileService is bypassed for these endpoints; tile download logic is duplicated.
  • Suggestion: Route all tile operations through ITileService.
  • Resolution (AZ-310 + AZ-311): ITileService extended with GetOrDownloadTileAsync and DownloadAndStoreSingleTileAsync; both endpoints now inject only ITileService. Caching (IMemoryCache), repository lookup, and downloader fallback consolidated inside TileService. Verified by 5 new unit tests + smoke integration.

F4: No physical boundary in Services project (Medium / Architecture) — RESOLVED

  • Location: SatelliteProvider.Services/ (all files)
  • Description: Three logical components (TileDownloader, RegionProcessing, RouteManagement) share one .csproj. No compiler-enforced boundary prevents direct cross-component coupling.
  • Impact: Over time, services may accumulate hidden coupling that's hard to detect without code review.
  • Suggestion: Accept as-is for current scale; consider splitting into separate projects if the codebase grows significantly.
  • Resolution (AZ-312 + AZ-313 + AZ-314, commit 8b0ddae): Project split into SatelliteProvider.Services.TileDownloader, SatelliteProvider.Services.RegionProcessing, SatelliteProvider.Services.RouteManagement. None of the three references another sibling — cross-sibling calls now flow exclusively through interfaces in SatelliteProvider.Common.Interfaces. RateLimitException relocated to SatelliteProvider.Common.Exceptions to keep the sibling boundary clean. Per-component DI extension methods (AddTileDownloader, AddRegionProcessing, AddRouteManagement) registered from Program.cs. Verified by 0 build errors, 40/40 unit tests, and full smoke integration suite passing post-split.

F5: module-layout.md vs DataAccess→Common dependency (Low / Architecture) — REVISED 2026-05-11

  • Location: _docs/02_document/module-layout.md
  • Original baseline claim: DataAccess was documented as "Imports from: Common" but the inspector concluded DataAccess had no ProjectReference to Common nor any using SatelliteProvider.Common.
  • Re-verified 2026-05-11 during 03-code-quality-refactoring (AZ-377): the original baseline scan was inaccurate. SatelliteProvider.DataAccess.csproj line 18 has a <ProjectReference Include="..\SatelliteProvider.Common\SatelliteProvider.Common.csproj" /> that has been present since at least the AZ-309 split, and 5 DataAccess files import SatelliteProvider.Common.Enums (RegionRepository.cs, IRegionRepository.cs, Models/RegionEntity.cs, Models/RoutePointEntity.cs, TypeHandlers/EnumStringTypeHandler.cs).
  • Impact: Documentation inaccuracy. Caused the AZ-315 doc-sync to "resolve" the finding by claiming DataAccess has no Common dep — that resolution itself was incorrect.
  • Resolution (AZ-377 + Phase 7 of 03-code-quality-refactoring run, commit 9a53bff): module-layout.md now correctly lists DataAccess ProjectReferences: SatelliteProvider.Common and enumerates the 7 import sites (5 enum + 1 MapConfig.DefaultTileSizePixels + 1 GeoUtils.*). The §Verification section now states the actual constraint: Common MUST NOT import from DataAccess (one-way leaf invariant). AZ-377 also widened the dependency surface from Common.Enums only to also include Common.Configs and Common.Utils — see batch 24 cumulative review F1.

Summary

Severity Count (baseline) Count (post AZ-309)
Critical 0 0
High 2 0
Medium 2 0
Low 1 0

The two High findings both relate to the same root cause: the ISatelliteDownloader abstraction was created but never wired into the system. The concrete GoogleMapsDownloaderV2 is used directly everywhere. This is the primary architecture gap — addressing it would enable the provider-agnostic design the system intends to have.

Post-AZ-309 Status

Epic AZ-309 (architecture coupling refactor) closes all five baseline findings:

  • F1, F2 — pre-AZ-309 cleanup (interface implementation + DI rewire).
  • F3 — AZ-310 + AZ-311 (route tile endpoints through ITileService).
  • F4 — AZ-312 + AZ-313 + AZ-314 (split monolithic Services csproj into three per-component csprojs with compiler-enforced sibling boundary).
  • F5 — AZ-315 (this commit; documentation now matches actual ProjectReference graph).

No new Architecture findings were introduced by the refactor. The baseline is now clean.

Reference in New Issue View Git Blame Copy Permalink