Some of the `update_osm_${LAYER}`-functions, which are executed by triggers on updates, execute an UPDATE statement on the same tables that have these triggers. This causes the trigger functions `flag` and `store` to run multiple times for one record.
For instance, if I add log in the trigger functions and run an INSERT on `osm_housenumber_point`, this output is generated:
```
NOTICE: Store
NOTICE: Flag
NOTICE: Refresh housenumber
NOTICE: Flag
NOTICE: Store
NOTICE: Flag
INSERT 0 1
```
If we limit the triggers from executing if they are called from another trigger using `pg_trigger_depth() < 1`, the triggers will only be triggered once per record:
```
NOTICE: Store
NOTICE: Flag
NOTICE: Refresh housenumber
INSERT 0 1
```
This will prevent redunant executions and might improve update performance.
Co-authored-by: Patrik Sylve <patrik.sylve@t-kartor.com>
I noticed slow performance in `getmvt` when generating tiles at zoom levels ~ 0-8.
The issue was due to CASE blocks in the `transportation` and `poi` layers, which were processing unneccessary rows at lower zoom levels. To optimize this I added a pre-filter on `zoom_level` to reduce the workload.
In my tests, I timed the query `select getmvt(0,0,0);`.
`area=europe/united-kingdom`:
* Before: ~17 seconds
* With this fix: ~80 ms
`area=planet`:
* Before: have not yet timed it, but i believe it was around 15min if i remember correctly
* With this fix: ~10 seconds
Co-authored-by: Patrik Sylve <patrik.sylve@t-kartor.com>
Fixes#1640
This ensures that place and poi points computed from areas result in a point node that is actually inside the area that it's derived from. Unusual shaped areas (like a banana shape, for example) may produce points that aren't actually contained.
The main goal is to improve consistency and reduce the number of poi without an icon.
- Remove music from poi cultural, music has only two subclasses music and musical_instrument, both are styled as poi shop
- Remove arts_centre from poi shop, it's already styled as poi cultural
- Remove chocolate icon, chocolate uses confectionery icon (see: https://github.com/openmaptiles/openmaptiles/pull/1628)
- Remove references to ice_cream class from poi shop
- Add missing gallery icon and move it to poi cultural (see: https://wiki.openstreetmap.org/wiki/Tag:tourism%3Dgallery)
- Use the icons alcohol, newsagent, perfumery, trade respectively for wine, kiosk, perfume and wholesale.
- Add a generic fallback icon for shops
According to [OpenMapTiles Schema](https://openmaptiles.org/schema/) ice_cream should be a class, not a subclass of the shop, so I think it makes more sense to remove ice_cream from the shop class. I tested this, and after the change, both amenity=ice_cream and shop=ice_cream OSM tags get class=ice_cream + subclass=ice_cream, which is fine, because both OSM tags shop=ice_cream and amenity=ice_cream represents the same thing
The next thing I did was to change the color of the ice_cream icon.
This PR is a **documentation change only**. It documents the fact that `name:xx` values are present in the tiles and documents that `name_de` and `name_en` will be removed in a future release in favor of `name:de` and `name:en` respectively.
Fixes#1606
Unblocks ZeLonewolf/openstreetmap-americana#1022
This PR is a bugfix which ensures that `amenity=university` is encoded as `class=college` rather than `class=office`. It's required because of a subclass collision for `university`, which exists under both class values.
Omitted values include:
* office=yes ... too generic
* office=vacant ... not really a POI
* deprecated values (like office=administrative)
* tags that don't have a clear meaning (like office=camping,
office=healthcare, office=transport, office=design)
* tags where another tag is used more often, e.g:
office=religion is used more often than office=parish
office=construction_company is used more often than office=construction
office=engineer is used more often than office=engineering
---------
Co-authored-by: Tomas Pohanka <TomPohys@gmail.com>
Paint shop is missing from the schema, so shops of this type do not appear on the map beside other POIs
---------
Co-authored-by: Tomas Pohanka <TomPohys@gmail.com>
Improved update performance of poi layer
- Refactored update_poi_point.sql to partial diff update
- Refactored IDs to be unique in poi_polygon.osm_ids
- Restricted updates to INSERT and UPDATE operations during poi_polygon.refresh
- Added analyze statements before update queries during poi_polygon.refresh
This PR adapts the code used in #1457 to allow POIs to render up to z10 for very large features. I set the threshold at a very conservative value of 10% of a tile. Thus, it will render if an area POI covers at least 10% of a tile at a given zoom. Only for 'university' and 'college'.
Additionally, this consolidates a double UNION ALL on the same table with a WHERE/CASE combination, which is simpler.
This PR addresses that by:
* rendering oneway arrows on more than `subclass=cycleway`
* less opacity for path/footway casing
* adding the thin black casing to bridges
* adding `subclass=steps` & `subclass=bridleway` rendering
* render `bicycle=designated` as `subclass=cycleway` and `horse=designated` as `subclass=bridleway` (bike taking priority)
* improved tunnel rendering with double casing & adding main line
* made casing min zoom consistent across subclasses
All of these changes were made to mimic OSM Carto.
* OpenMapTiles style
* Rename style snippets from layer_name.json to defaultstyle.json
* Add README to icons dir.
* Add README to icons dir.
* Rename snippets to style.json
* Fix spritezero - move icons to svg subdir
* Move icons from style/icons/svg directly to style/icons
Co-authored-by: Tomas Pohanka <TomPohys@gmail.com>
closes#1370
This PR adds a parcel locker to post class.
Similarly to ATMs from PR #1375 if a name is not present we try to use `brand` or `operator`. Here we also try to add `ref` to the brand or operator since it can disambiguate parcel lockers that are next to each other (which happens).
After each diff import the table `osm_poi_point` is fully rewritten due to a full update of the field tags. It is now good to do an a update, event if the content does not change, postgres delete and reinsert internally the record. Resulting in more write and internal table size raising.
Note: not directly linked, but there is a problem in this case, the autocaccum is not sufficient to keep this table size moderate, but grow indefinitely.
Replacing update on the whole table with an update only on changed rows.
The goal is to update more quickly by just updating the changing content.
The update now focus on osm_id of changed rows, it use index. Add a where clause tags != update_tags(tags, geometry) en ensure only update when changed.
It requires one more trigger and a table to store changed osm_id.
The UPDATE is keep in a function to be reusable for initial setup and trigger update.
I try many code layout before done it in this way with the goal to keep the code for initial pass and for update. It should have low impact on initial data load. Better performance for row update can be achieve with BEFORE UPDATE, but require to duplicate the logic.
It is not based on the already merged https://github.com/openmaptiles/openmaptiles/pull/896 because calling and update within a function for each updated row was not efficient for larger table (like housenumber).
It addresses #814.
* Remake update_peak_point use incremental update #814
* Make update_aerodrome_label_point use incremental update #814
* Make housenumber_centroid use incremental update #814
* Make update_continent_point use incremental update #814
* Make update_island_point use incremental update #814
* Make update_island_polygon use incremental update #814
* Remove dead code in update_state_point.sql
* Make update_state_point use incremental update #814
* Remove dead code in update_country_point.sql
* Make update_country_point use incremental update #814
* Make osm_poi_polygon use incremental update #814
Thanks @frodrigo
