Interview with Developer Jose Luis Arenas regarding Google App Engine & Geospatial Datas
2 likes2,321 views
The document details an event organized by GDG London focused on Google Cloud Platform, including a retrospective game, developer interviews, and a hackathon using Open Data. The hackathon is scheduled for July 18-20 and invites participants to build apps in 48 hours with datasets from Data.gov.uk. Additionally, it discusses application architecture challenges regarding Google Cloud APIs and features feedback mechanisms for event attendees.
![Why not do this in datastore?
Inequality Filters Are Allowed On One Property Only
Street.query().filter(‘lat >’, min_lat).filter(‘lat <’ max_lat).filter(‘lon >’, min_lon).filter(‘lon <’,max_lon)
Workarround:
[s for s Street.query().filter(‘lat >’, min_lat).filter(‘lat <’ max_lat) if s.lon > and s.lon < max_lon]
but an stripe of 2 km wide from East to West in UK has lots of streets….](https://image.slidesharecdn.com/interviewwithdeveloperjoseluisarenas-140501075340-phpapp02/85/Interview-with-Developer-Jose-Luis-Arenas-regarding-Google-App-Engine-Geospatial-Datas-18-320.jpg)
Interview with Developer Jose Luis Arenas regarding Google App Engine & Geospatial Datas
- 1. GDG London & Google Developer Experts at Google London
- 2. Next 2 hours ● Who is GDG London & GDE ● What we trying to do ● Retrospective Game on Google Cloud Platform ● Interview with Developer Jose Luis Arenas ● Retrospective Game continue ● Wrap up ● 8:15pm Live Broadcast After Party
- 3. Goal of the Group ● Hold event every 2 months ● Work with PHP, Java, Python and other user groups ● Spread the knowledge of Google Cloud Platform ● Give feedback to Google
- 4. Where to find us ● Find out about our events & goals at http: //www.meetup.com/Google-Cloud-Platform- App-Engine-Compute-Engine-BigQuery/
- 5. Google Cloud Platform Hackathon Building apps with opendata from data.gov.uk ● Friday 18th July to 20th ● Build apps in 48 hours using data from data. gov.uk ● 13,000+ government dataset ● Hosted at Google Campus
- 6. Retrospective Game Write on coloured Post-it note your ideas for the following. Only 1 idea per Post-it note Green - Favourite announcement in the keynote Pink - Feature Request for Google App Engine Orange - Feature Request for Google Compute Engine Yellow - Other Feature Other info Job Role & Which product you're using.
- 7. Interview with Developer Jose Luis Arenas Jose Luis Arenas from NDS & TeamUno Rif Kiamil from FOODit & TeamUno
- 8. Problem Need to find all the names of streets in a radius. Wanted to take open source data “OS Locator” from Ordnance Survey and make it into an API.
- 10. Application Architecture v1 Search API is the forgotten database of Google Cloud Platform. It has it’s own SLA and totally different to datastore
- 11. Search API still learning to count Search API - Giving us results outside search radius Link to source code https://github.com/FOODit/street-geo-search/blob/master/streetsearch.py#L123 Issue 8824: Distance queries in Search API return results that are further away than specified https: //code.google.com/p/googleappengine/issues/detail?id=8824
- 12. We could have ● Used CloudSQL - Cost a lot more! Customer said No (see mySQL geo euery example) ● Tried GeoHashing with a Project called “geomodel” - But it was too slow. But we liked to idea of it using datastore over search api. (see geomodel example)
- 13. Went back to Search API, but with our workaround
- 14. Latitude
- 16. Check which 4 results are 3km away
- 17. Using Haversine Formula found objects
- 18. Why not do this in datastore? Inequality Filters Are Allowed On One Property Only Street.query().filter(‘lat >’, min_lat).filter(‘lat <’ max_lat).filter(‘lon >’, min_lon).filter(‘lon <’,max_lon) Workarround: [s for s Street.query().filter(‘lat >’, min_lat).filter(‘lat <’ max_lat) if s.lon > and s.lon < max_lon] but an stripe of 2 km wide from East to West in UK has lots of streets….
- 20. End Points Go to https://street-geo-search.appspot.com/
- 21. Why I love endpoints Google APIs Explorer Field filtering and versioning Generate client libraries for iOS and Android
- 22. End Points
- 25. What I like about End Points & GAE I like this style as never have to think about it again.. I know always going to work and not going to cost me or client much!
- 26. I liked Endpoints so much that.. My new startup TeamUno using just Google End Points and AngularJS We don’t have any server side templates. I would love to see it Google make it easier to bind datastore/cloudsql to End Points ( LESS CODE)
- 27. Find us online at Jose Luis Arenas jla@sblk.co from NDS & TeamUno Rif Kiamil rif@foodit.com from FOODit & TeamUno
- 28. Notes - Import How did you do the import… https://github.com/FOODit/street-geo- search/blob/master/tools/OSLocator-Import.php How did covert OS Grid References to Lat & Long PHPCoord http://www.jstott.me.uk/phpcoord/
- 29. Notes - mySQL Geo Query $query = <<<EOD SELECT NAME, CLASSIFICATION, SETTLEMENT, LOCALITY, COU_UNIT, LOCAL_AUTHORITY, TILE_10K, TILE_25K, SOURCE, CENTLAT, CENTLNG, DISTANCE FROM ( SELECT NAME, CLASSIFICATION, SETTLEMENT, LOCALITY, COU_UNIT, LOCAL_AUTHORITY, TILE_10K, TILE_25K, SOURCE, CENTLAT, CENTLNG, r, (6378.10 * ACOS(COS(RADIANS(latpoint)) * COS(RADIANS(CENTLAT)) * COS(RADIANS(longpoint) - RADIANS(CENTLNG)) + SIN(RADIANS(latpoint)) * SIN(RADIANS(CENTLAT)))) AS DISTANCE FROM OSL JOIN ( SELECT $lat AS latpoint, $lng AS longpoint, $rad AS r ) AS p WHERE CENTLAT BETWEEN latpoint - (r / 111.045) AND latpoint + (r / 111.045) AND CENTLNG BETWEEN longpoint - (r / (111.045 * COS(RADIANS(latpoint)))) AND longpoint + (r / (111.045 * COS(RADIANS(latpoint)))) ) d WHERE DISTANCE <= r ORDER BY DISTANCE EOD;
- 30. Notes - Example of GeoHashing geomodel Project https://code.google.com/p/geomodel/ from geo.geomodel import GeoModel class OSLocator(GeoModel, ndb.Model): ……. entity = OSLocator() entity.location = ndb.GeoPt(item.centlat, item.centlng) …… results = OSLocator.proximity_fetch( OSLocator.query(), ndb.GeoPt(request.latitude, request.longitude), max_results=10000, max_distance=int(request.radius * 1000.0))
- 32. Retrospective Game Write on coloured Post-it note your ideas for the following. Only 1 idea per Post-it note Green - Favourite announcement in the keynote Pink - Feature Request for Google App Engine Orange - Feature Request for Google Compute Engine Yellow - Other Feature Other info Job Role & Which product you're using.