Some features disappeared after the refactoring, so this at least makes the where clause match the case clause entries. Fixes #364
…ap-carto into math1985-roadsv5cnew
Code for every highway type is currently spread out over three places in the code: once for bridges, once for tunnels, and once for normal roads. This means that a lot of code is duplicated. The current pull request brings the code for a single highway type together at one place in the code. Although the code is brought together, it does not yet collapse all equivalent definitions. This will be done in the next pull request. The advantage of this is easier maintenance, because a) the CartoCSS code will be shorter; b) there will be less duplication in the code; and c) code for one type of road is at only one location in the code. Changes ------- Although the aim of this change is to improve the structure of the code, it has some small changes of rendering as side effects: - The tag access=no is no longer rendered on railways, aeroways, and roads under construction. - For bridges and tunnels, access restrictions are rendered at the same time as the roads to which they apply, just like it was already the case for normal roads. - Direction arrows are now layered as own attachment, but still directly on top of the layer they apply to. - Tunnels and bridges that are a link are now rendered under other roads of the same osm-layer, just like links that are not a tunnel or bridge. - The following kind of tunnels are now correctly rendered in the tunnel layer instead of in the regular layer: the highway types tertiary (z10-12), residential (z10-12), unclassified (z10-12), service, living_street, road (fill only), road (z10-12), pedestrian, platform, steps, track, raceway, construction, proposed; all railway types; all aeroway types. - The following kind of bridges are now correctly rendered in the bridge layer instead of in the regular layer: the highway types secondary (z12), tertiary (z10-13), unclassified (z10-13), residential (z10-13), road (z10-13), living_street, pedestrian (z13-15), bridleway (z13), footway (z13), cycleway (z13), path (z13), track (z13), raceway, construction, platform, steps; the railway types spur (z11-12), siding (z11-12), yard (z11-12), funicular, minitiature, tram, lightrail (z13), subway (z12-13), preserved, monorail, platform, turntable; the aeroway type runway (z11-13). - Ways with both tunnel=yes and bridge=yes might get rendered different than before (but such ways should not exist anyway). - highway=road is now also rendered on the layer for bridges and tunnels. - One-way restrictions on bridges are no longer displayed on highways types that are not rendered. - One-way restrictions on bridges are now displayed on all rendered railway types. Performance ----------- In general performance seems to improve. I compared performance between 2254ec9 and 42e4203. The resulting output osm.xml is now 8% smaller. I measure a performance loss of 4% at z15, but performance gains of 12%, 16%, and 13% on z16, z17, and z18, respectively. My measurements are not very accurate, so I would be happy if @pnorman or someone else could independently verify this. Pull request size ----------------- I'm sorry for the huge pull request. I couldn't break it up without breaking functionality or significantly increasing run-time. The next pull requests will be smaller.
It's been recommended by a native Khmer speaker, and is packaged on Ubuntu. Also split the font lists into multiple lines because it's getting long
…s for languages you care about are required
Padauk has both a normal and a bold version, making it a good choice. It's also the only packaged Burmese font I could find.
Arundina Sans is based on DejaVu and has bold and italic versions, so it's an easy choice. Obtained with the ``fonts-sipa-arundina`` package. This change makes mapnik/mapnik#2161 more obvious, but we were probably encountering that bug before with Unifont, it's just hard to tell with how hard Unifont was to read.
Feedback is that Droid Fallback looks better than Unifont. As Droid Fallback doesn't have bold, hopefully we don't get this far, but it beats Unifont. The droid fonts can be found in the fonts-droid package on Ubuntu. A readme section to list required font packages has also been added
…-carto into mrwojo-area-labels
Fix historic labels
This reverts commit b0842bd. Refs #348 #344 #348 (comment) If this is making some boundaries disappear, then this clearly needs a different approach or much more discussion.
Effectively a rebase of #348. This isn't ideal, since we should be able to (somehow) figure out the max(admin_level) from all the ways and relations automatically, but the osm2pgsql database layout doesn't make it easy. Since mappers are commonly doing this algorithm anyway by hand - well, hey-ho. It's still an improvement in the end result.
… the interior.
… into carto.
Previously it was very eye-catching blue dashes, which was also easily confused with cycle paths. The gradient help indicate which side of the line is 'in' the marina, without colouring in the whole thing.
Ideally the labels would be offset to indicate the area covered, but boundaries aren't in the polygon table.
…ildings. There are probably other areas labels worth adding back in too.
This removes them from unrendered highway values, and also removes them from railways and waterways where they are of dubious interest and poor colour choice respectively.
We should be ignoring the '1' and 'true' versions, and beyond 'yes' and '-1' we don't actually render anything.
This used to pick up lots of non-rendered features, as well as a few others that will need to be re-instated elsewhere (e.g. admin boundaries).