fix scalerank 0 region boundaries, drop low-zoom names, add map_unit

[nvkelso]

connects to #992 to fix scalerank 0 region boundaries
also adds map_unit boundaries and
drops name at low zooms 0,1,2,3,4,5,6

• Update tests
• Update data download and config
• Update docs

[nvkelso]

@zerebubuth Looks like my adding map_unit as a new Natural Earth data source is making all the tests fail. Can you take a look, please?

[zerebubuth]

The first error seems to be:

ERROR:  column "mapcolor9" of relation "ne_10m_admin_0_boundary_lines_map_units" does not exist
invalid command \N
invalid command \.
ERROR:  syntax error at or near "5.00000000000"
LINE 1: 5.00000000000 Map unit boundary 
        ^
ERROR:  current transaction is aborted, commands ignored until end of transaction block

Which is preventing the shapefiles for the map units being loaded. It looks like the schema doesn't contain a column called "mapcolor9", but the data does. We don't currently support loading a subset of the columns - they must all be specified in the schema.

The second error seems to be:

psql:apply-updates-non-planet-tables.sql:107: ERROR:  column "scalerank" does not exist
LINE 15: ...recla" = 'Admin-1 region boundary' THEN CASE WHEN scalerank ...

Which probably means that extra_columns: scalerank needs to be specified when using the global ne_region_boundaries_min_zoom.

[nvkelso]

Thanks! Fixed in b39f38b and 26e1c15, waiting for CirlceCI to confirm.

[rmarianski]

no trailing newline

[nvkelso]

Fixed in ad617c9.

[rmarianski]

LGTM

[rmarianski]

I checked in some fixes. We'll see if the tests pass on circle now :fingers-crossed:

[zerebubuth]

We'll see if the tests pass on circle now

Do they pass locally? 😜

[rmarianski]

Do they pass locally? 😜

I'm going to wait until the circle build finishes before answering that 😝

[nvkelso]

Our current thinking is the data files here need to be in web Mercator (but are probably in geographic now).

[nvkelso]

Last test failures are related to OSM zoom 8 content, what gives?

[zerebubuth]

Boundaries looked like they just needed the names swapped, possibly because the direction was reversed due to the introduction of stable ordering in #1022.

The second failure looks like it was also a copy-paste issue with the tile coordinates.

Works for me now, let's see if CircleCI agrees 😉

[nvkelso]

Woot! Merging.