Hacker News new | past | comments | ask | show | jobs | submit login
Using OpenStreetMap and GraphHopper to visit every pub in Dublin (graphhopper.com)
85 points by rmc on Nov 19, 2015 | hide | past | web | favorite | 14 comments

I think 'every pub' is a bit of a stretch, I've seen a good few missing just from a quick glance at the map.

Apart from that, I do love these type of projects. People who see a story of a guy following something manual (a book) and thinking "Hmmm, I wonder if I could make something to help with that/do that for me"

It may not have counted "bars" - that is, wine bars, concert venues, hotels and the like.

However if there are pubs you know are still open (they've been closing at a fast rate) which are not in OpenStreetMap.org then do please add them yourself, or let someone know. The map is made up from people like yourself who know their local area.

edits - Running the same Overpass Query gets 255 "pubs".

http://overpass-turbo.eu/s/cP1 (132 areas for hotel, bar, pub)

http://overpass-turbo.eu/s/cP2 (219 points)

Total for pubs, bars, hotel: 351

I do contribute to OSM actually, really good project. Stuff like The Dawson Lounge and Porterhouse Central seem to be missing on their map which are present on OSM and OverPass (Dawson Lounge actually isn't on overpass).

Also, it seems they counted DTwo, which is more of a 'bar'.

It's the tagging, the article sticks to "amenity=pub" for simplicity, but OpenStreetMap might tag a place that serves alcohol "amenity=bar", "amenity=biergarten", "amenity=nightclub" or even "amenity=restaurant".

Some of that is driven by the data model supporting arbitrary key value pairs instead of being structured, but I think there is also a tendency to try to build a convenient taxonomy that has handy mental "bundles" instead of abstracting things a little bit.

A more abstract model might call them all amenity=meeting_place and then have flags for seating, alcohol, food, a dance floor, outdoor seating, etc., and leave the finer categorization up to the end consumer of the data.

(I realize someone fooling around with OSM would know this, I'm trying to write for the HN audience)

The data is from OpenStreetMap, so anyone can add them and improve the map.

Which pubs are missing? I can add them myself.

Porterhouse seems to be missing but it's on OSM and Overpass Turbo. The Dawson Lounge doesn't seem to be on the map either

Also DTwo is there yet Dicey's and Coppers are not, which seems odd (Krystle is on Overpass too), so maybe it's just the definition of 'pub' used that's throwing it a bit e.g. Dandelion isn't listed, yet it seems Capitol is. That could just be due to the small map though and I'm reading it wrong (Maybe Capitol is JJ's actually)

I think in this case they had the route optimization software and saw a nice opportunity to illustrate what it can do.

True, and it is a nice demonstration. Show's the real-world applicable use of it.

Then add them to the map? :-)

Peter, one of the people behind GraphHopper, here. Let me know if you have any questions

You could just miss out Temple Bar too, unless you want tourist traps galore.

Challenge Accepted! j/k

This is a great example/demonstration of graph hopper (which I had never heard of before). Traveling salesmen problems are always incredibly challenging.

Just a side note: the GraphHopper Routing engine is open source: https://github.com/graphhopper/graphhopper/ but is not able to solve TSP or vehicle routing. For that we've developed jsprit, also open source: http://jsprit.github.io/

And both combined to be easily used and very fast in our commercial offering called the GraphHopper Directions API

27 days eh...? Temptation rising...

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact