Search for also places that correspond to addr:street

[johsin18]

This is a PR to solve issue #111.

All commits but the last are cleanup and refactoring.

I tested only using osmi-testzone.osm, and I ran it once on the Swiss data set. I was not able to test the results graphically (probably involves some huge setup). Can somebody please test it graphically, e.g. the addresses around Tellplatz Basel should not be shown red anymore: http://tools.geofabrik.de/osmi/?view=addresses&lon=7.59483&lat=47.54331&zoom=17&overlays=buildings,buildings_with_addresses,postal_code,entrances_deprecated,entrances,street_not_found,interpolation,interpolation_errors,connection_lines,nearest_points,nearest_roads,nearest_areas,addrx_on_nonclosed_way

[johsin18]

@ltog @Nakaner I would appreciate your comments.

[ltog]

@johsin18 (CC @Nakaner ): Thank you for taking the time to write the pull request and sorry for the delayed answer. I started working on a file for Vagrant to resolve your mentioned

I was not able to test the results graphically (probably involves some huge setup).

by providing a one-click solution to setup osmi-addresses. While preparing the Vagrantfile I discovered some issues with the GDAL version provided by Ubuntu 18.04 which I began to address here: https://github.com/ltog/osmi-addresses/commits/gdal_fix

I'll be at SotM next weekend. If you'll be there too, we could meet.

[johsin18]

So finally I was able to set up MapServer and test my changes on the map. Following the README, it was less hard than expected. I though I would have to also set up a server serving the base tile layers, but that's not needed, fortunately.

So indeed my code was still buggy, but that should be fixed now.

I would appreciate if you would merge my branch.

[ltog]

@johsin18 : I'm glad the README was helpful...

Recently I met with @Nakaner . We weren't sure if tying together addr:street=* with place=square, name=* is something we should encourage through osmi-addresses. We therefore decided to ask for a broader opinion at the tagging mailinglist. You can find my message at https://lists.openstreetmap.org/pipermail/tagging/2018-August/038415.html . The whole structure of the discussion will be visible at https://lists.openstreetmap.org/pipermail/tagging/2018-August/thread.html once it took place.

Sorry, that this takes so long...

[johsin18]

Well, you had almost one year time to think about this, and now you change your minds :-(

If you don't want this change, please merge at least my code cleanups, so that my work was not fully in vain.

[Nakaner]

Thank you for your contribution.

I decided to merge this pull request (not the refactoring only) because there are indeed cases where place=square looks unavoidable to me. The Tellplatz example in Basel is not such a case because it would be possible to name the pedestrian areas.

[ltog]

@Nakaner : Thank you for taking care of this!

[Nakaner]

The Mapserver styles have been deployed now.

[johsin18]

Thanks for taking this in.