• Home
  • Features
  • Pricing
  • Docs
  • Announcements
  • Sign In

MapColonies / shigola / cbbcd55e6
44%

Build:
DEFAULT BRANCH: master
Ran 30 Aug 2026 07:01PM UTC
Jobs 1
Files 198
Run time 1min
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst

30 Aug 2026 06:58PM UTC coverage: 46.759% (-0.7%) from 47.414%
cbbcd55e6

push

github

web-flow
test(tilecontent): verify tile contents end to end (MAPCO-11545) (#12)

* test(mvt): add a shared way to decode and render a tile in tests

More than one test needs to look inside a vector tile, and they should describe
what they find the same way. provider/postgis asserts what the database
encoded; the tile-content checks assert what a client received. Two consumers
is a real seam rather than a hypothetical one, so the vocabulary is extracted
before the second one is written rather than after.

Two things it does that a test would otherwise redo badly. It resolves the
specification's delta-encoded geometry into absolute tile-space positions,
because a test wants to say where a feature is rather than how far it moved
from the last one. And it orders layers by name, because the specification does
not order them within a tile: anything reading them positionally pins something
ST_AsMVT is free to change.

AssertGolden carries the -update-golden flag and, in its comment, the reason a
golden file alone is not enough. The way a golden fails is that someone
regenerates it, skims a diff they cannot read, and commits; what stops that is
the fixture staying small and the test asserting some facts literally, and
neither of those can live in this package.

Groundwork for MAPCO-11546.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* test(server): check tile contents where the two schemes disagree

A black-box check that puts known rows in PostGIS, asks the running server for
the tiles covering them over real HTTP, and asserts what comes back down to
exact tile-space coordinates.

Nothing did this before. TestMVTProviders decodes a tile but calls the provider
directly and never crosses HTTP; the OGC CITE suite fetches real tiles but
checks conformance rather than content, and its runner says outright that it
passes against an empty tile. Everything between the provider and the socket --
OGC path parsing, the row-before-column transposition, gz... (continued)

0 of 276 new or added lines in 1 file covered. (0.0%)

9342 of 19979 relevant lines covered (46.76%)

207.19 hits per line

Uncovered Changes

Lines Coverage ∆ File
276
0.0
internal/mvttest/mvttest.go
Jobs
ID Job ID Ran Files Coverage
1 cbbcd55e6.1 30 Aug 2026 07:01PM UTC 198
46.76
GitHub Action Run
Source Files on build cbbcd55e6
  • Tree
  • List 198
  • Changed 0
  • Source Changed 0
  • Coverage Changed 0
Coverage ∆ File Lines Relevant Covered Missed Hits/Line
  • Back to Repo
  • cbbcd55e on github
  • Prev Build on master (#08F1E3958)
  • Next Build on master (#AEB759B15)
  • Delete
STATUS · Troubleshooting · Open an Issue · Sales · Support · CAREERS · ENTERPRISE · START FREE TRIAL · SCHEDULE DEMO
ANNOUNCEMENTS · TWITTER · TOS & SLA · Supported CI Services · What's a CI service? · Automated Testing

© 2026 Coveralls, Inc