MapColonies / shigola / 2bc2be592
48%

DEFAULT BRANCH: master
Ran
Jobs 1
Files 203
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
coverage: 47.681% (+0.02%) from 47.662%
2bc2be592

push

github

web-flow 
test(postgis): carry athens in the fixture and rename it to shigola (#5)

* test(postgis): carry athens in the fixture and rename it to shigola

The fixture database name lives inside the dump, not in references to it: a
pg_dump custom-format archive records the database it was dumped from, and
pg_restore -C recreates that name. Renaming every mention of "tegola" in the
tree therefore never renamed the database. Regenerating the archive from a
database that is genuinely called "shigola" is the only fix, and it is the one
moment the Athens OSM extract can be added at no extra cost.

The three Athens layers .github/cite/config.toml serves -- land_polygons,
roads_lines, places_points -- are now in the fixture, so the data the OGC
conformance suite exercises is reachable through mvt_postgis and not only
through the GeoPackage provider. TestMVTProviders covers them on the same tile
.github/workflows/ogc_cite.yml passes to run.sh, so a tile chosen independently
cannot drift into empty ground and turn "has no features" into a false alarm.

generate-dump.sh rebuilds the archive from the previous dump plus the
GeoPackage, needing only Docker. Both container images are pinned, GDAL by
digest: ogr2ogr decides the shape of the Athens tables, so a floating tag would
make the output depend on the day it ran. Regeneration is reproducible in
content, not in bytes -- pg_dump stamps a creation time into the header.

migration.sh drops the pre-rename database and role rather than ignoring them,
so a volume predating the rename cannot keep a stale, Athens-less "tegola"
alongside the real one for PGURI to find. It also gains pipefail: every command
in it runs through run(), which pipes into sed, and a pipeline's status is its
last command's -- so a failed pg_restore was reported as sed's success and the
script printed "Migration completed successfully" over an empty database.

CI now blocks on the migration container instead of assuming `up -d` means the
fixture is loaded... (continued)

9614 of 20163 relevant lines covered (47.68%)

205.94 hits per line

Jobs
ID Job ID Ran Files Coverage
1 2bc2be592.1 203
47.68
 GitHub Action Run
Source Files on build 2bc2be592