OpenStreetMap

SomeoneElse has commented on the following diary entries

Post When Comment
Trees (again) 5 days ago

@BushmanK Interesting - I've often added details conveying that sort of information as part of a note (see e.g. http://www.openstreetmap.org/way/343924077 ) - maybe "wood:age" et al is worth looking at.

Validating the map - Part 1 22 days ago

As well as "detecting problems" we also need to ask why a problematic edit was made.

For example, "doodle the dog" in the example above - was that drawn by a bored college student who'd been dragged into a HOT mapping session against their will? If so, perhaps we could try and engage with the person running the class?

MAPS.ME is now an editor 25 days ago

We're seeing a definite spike in new editors near me in the UK, compared to (say) 6 days ago:

http://resultmaps.neis-one.org/newestosm?c=United%20Kingdom#8/52.955/-0.654

What's really interesting is that there's actually roughly twice as many new mappers using iD as last week, not even counting the new MAPS.ME ones! It'd be interesting to see how widespread that is.

Mapping missing motorway exit numbers 26 days ago

On countries that don't use junction reference numbers as much as some others (like the US), I wouldn't expect too much from a query looking for junction reference numbers; just the same as in the UK (where there aren't multiplexed routes) I wouldn't expect to be able to extract sensible and complete route relations.

Just because the wiki says to do something it doesn't mean it's commonly done or even a good idea; sometimes it represents only the view of the person who last updated that wiki page.

In the US I'd spend time on misaligned and mismapped TIGER data before even thinking about whether a nonexistant junction reference number is indicated by "noref=yes" :)

"This is too complicated" - is it? 26 days ago

It's only "too complicated" for a new user if they aren't guided through the process when they choose to add a диспансер. New users don't read the wiki and don't read the email lists or the forum (initially they probably don't know that any of these exist). What new users do see on sign is:

The CTs, which the have to agree to. A confirmation email, which they need to click on a link on. "http://www.openstreetmap.org/welcome", which has a big "Start Mapping" button on it which they'll probably click on*. An osm.org screen centred at their browser location If they click "Edit" they're offered the choice of the "iD Walkthrough" or to start editing immediately.

iD (at least in my English locale) doesn't know what a "диспансер" is and if they search for it and there are no local matches the first worldwide match is a hospital in Serbia. Compare that with adding a feature that iD does know about.

  • The "welcome" does have a link to http://www.openstreetmap.org/help , which does link to external resources such as the wiki, but it's some way down in the list of things presented to newcomers, and not really understandable by them when they get there.

To summarise, if you want people (especially new users) to add these tags, you'll need to make it easy for people to add them. Simply writing them down somewhere that new users won't read doesn't do that.

As an aside, you might like to know that "healthcare" is the current quarterly project in GB. See this post on the GB list https://lists.openstreetmap.org/pipermail/talk-gb/2016-March/018720.html .

Sidewalks! about 1 month ago

Actually, I think the left/right thing could probably be handled by "line-offset", as on the example here:

http://gis.stackexchange.com/questions/128244/tilemill-inset-line-with-negative-line-offset-property-value-produces-strange-a

However I don't think I'll worry about left vs right sidewalks (or at least not yet). I'd want to remove some of the duplication I've introduced into roads.mss first.

The "other" problem (where an unclassified road joined a tertiary it "poked through" and could be seen over the casing at the other side of the road was fixed by setting "line-cap: butt" instead of "round".

African Roads and a Western Bias in Mapping about 1 month ago

You're absolutely right - the highway tags in OSM do have a Western bias (strictly speaking, a Western European bias, since that was where the tags originated - there are parts of the US where they don't work well, either).

However, as Sanderd17 mentioned above, there's nothing to stop people from adding more detail - things like surface, tracktype, etc.

It's also possible, when you're creating maps for your Nuvi, to process different sorts of data in different ways. For example, when I create maps for mine I have it ignore roads and tracks that are probably private. There are a few questions over at https://help.openstreetmap.org/ that might help with that. Although it's more complicated than just downloading a map from someone else, it's not that difficult.

If you get stuck with anything, just ask - there's a list of ways to get in touch on http://wiki.openstreetmap.org/wiki/Contact_channels that should help.

Sidewalks! about 2 months ago

@Stereo I'm (mis)using the Mapnik "casing" to show a "sidewalk" on both sides of the road. I've no idea whether it's possible to have a casing on only one side of a way, and if so how hard to do it would be. The "lua" side is pretty straightforward - instead of "tertiary_sidewalk" and "tertiary" you'd have "tertiary_left_sidewalk", "tertiary_right_sidewalk" and "tertiary". Of course, the more important information is "you can safely walk here" rather than "and it's on the left as you go north".

The router I use most often is the built-in Garmin eTrex one. Ages ago I did have a go at producing a Garmin map for "foot only use" (on it motorways weren't usable roads, for example), using a version of the C# preprocessor that I used already. It sort-of worked but wasn't ideal, since you couldn't use the same .img for e.g. car routing too.

I've not looked at OSRM et al, but as I understand it if you run your own instance of that it's highly configurable (see https://github.com/Project-OSRM/osrm-backend/wiki/Profiles ). It wouldn't surprise me if you could persuade it (or some other online router) to support sidewalk tags.

Showing off surface tags about 2 months ago

Re the "getting it online" bit - a quick and dirty Leaflet/Mapnik approach would be to use a lua script together with the existing "OSM Carto" style or a variation of it to display the data using existing tags.

https://github.com/SomeoneElseOSM/designation-style is an example of that, not with "paved", but with "designation" (which is used in England and Wales to indicate specific kinds of public access). It dates from when "highway=path" and "highway=footway" were rendered differently by "OSM Carto" - paths and footways with a designation are rendered as "footway", and paths and footways without as "path".

If you read through the github issues for the standard style there are a lot of comments saying "we can't do that because XYZ key isn't in the rendering database". Using lua avoids that problem; it can act on any key/value combination in the data being loaded.

Removed about 2 months ago

All I can suggest is to comment on the discussions on the changesets concerned - explain what the problem is, what they did wrong, and how you'd have done it. Say that you're local and offer to check anything they want the next time you're in the area of (whatever it is).

If you can and if appropriate, add changeset discussion comments in a language that the author is likely to speak as well as yours (even a Bing Transator or similar translated text is better than nothing).

If they don't reply after a week or so, explain again that it'd be really nice if they did reply, pointing out if necessary that you're another OSM mapper who just happens to live in the area; you're not some disembodied "system error message".

If that doesn't work and you still have a problem contact the Data Working Group at data@osmfoundation.org (disclaimer: I'm a member) and we can try suggest that they really ought to engage with you, and can take further steps if necessary.

Obviously I've no idea what the "project team" was that prompted you to write this diary entry, but I suspect that "team mapping" will be more prevalent not less as we go forward. Several companies who widely use OSM data have mapping teams, and there are lots of other "mapping projects" using OSM data. The quality of mapping from people on these teams varies, just like it does from other OSM mappers. Some "project team" mappers have been OSM contributors of long standing before starting their current team membership; others seem to have no knowledge of OSM and seem to have been picked up via an ad-hoc jobs board. In most cases the wider OSM community gets them "trained" eventually. The most common mistake "team leaders" (or whatever sort) make is forgetting that OSM is a community more than it is a project. If you come in and add any old rubbish to the data people will comment on it and fix it.

Myth of Newbie 3 months ago

"And here goes John Smith, who indirectly claims, that he knows how OSM newbies think and so on, when he refers to them. No, he knows almost nothing. "

So you're assuming that people making statements about what does and doesn't work for new mappers haven't ever investigated what works or not?

Do you have any evidence for that assumption?

Sidewalk tags in Cluj-Napoca (Cluj county, RO) - IN PROGRESS 4 months ago

Excellent stuff! As someone who regularly walks between towns when the weather doesn't really allow the use of footpaths etc., I find mapped footpaths far more useful than addresses :)

Re-tagging picnic sites with leisure=picnic_site and amenity=picnic_site to tourism=picnic_site (Part 2) 4 months ago

@jinalfoflia "last edited within a year" is about the threshold that I'd use for messaging previous mappers too.

Do you have any idea how many of the picnic sites in, say, the UK are as yet unmapped (there are now 1632 there according to http://taginfo.openstreetmap.org.uk/tags/?key=tourism&value=picnic_site , you changed 15 in https://www.openstreetmap.org/changeset/36125857 ).

Also how many of the things that are currently tagged as tourism=picnic_site do you think are mistagged?

Improving the OSM map - why don't we? [12] 4 months ago

I reckon that it's almost always possible to figure out what even "incomprehensible" tags actually mean (or at least what the mapper was trying to do at the time). A while back SimonPoole mentioned "yes=no": https://twitter.com/sp8962/status/643894726366789632 . I had a look at those, and even something with something as nonsensical as that it was possible to figure out the meaning (e.g. a mapper was trying to change tiger:reviewed=no to tiger:reviewed=yes). That's part of the reason why I find "tagfiddlers" (people who just remove tags that they don't understand, without asking the previous mapper) annoying.

On the more general point about source tags, if you really do use only one source for an entire changeset then it might make sense to use a changeset source tag (and when I do, that's what I do) but most of the time the source of anything is much more complicated (for me usually some combination of notes, new GPS traces, previous GPS traces, imagery and government open data). That's when I find element source tags useful.

Mapping a neighborhood park 4 months ago

Just to comment on the "better we upload all the data in english" reply...

As an English speaker, if I'm visiting somewhere where signposts primarily aren't in English (or are even in a non-Latin script) I'd actually prefer names on a map to match the signposts, rather than be a translation of the local name.

To take an example that I'm familiar with (in Sweden), http://www.openstreetmap.org/way/20278673 is referred to by everyone as the "Centralbron", and from memory it's signposted as that. In OSM someone's added a "name:en" of "Central Bridge" (that's just the English translation - I've never seen it signposted, although it was a while since I was there). In order for "Central Bridge" to be useful to me I'd actually need to be able to translate the English into Swedish, and then compare that with the signposts. That's actually more work than just having what's on the signposts on the map, even if I don't speak Swedish.

About Huts 4 months ago

The bit I don't understand is "... So, I must go along with the community". I don't understand what benefit there is to using an artificially circular way as opposed to a node (at least if way_area isn't going to be used to determine size - which if the examples on this page are anything to go by, it couldn't be reliably). Surely (while there are still things to be map) the benefits of mapping throughput would outweigh the prettiness of overnoded ways?

Private Plugin for "Faint" Trails? 5 months ago

I'm sure that there are map styles (if not public map tiles) that take trail_visibility into account. For my own use I maintain a map style designed for hiking in England and Wales, and that drops non-designated(1) low-visibility highways:

https://github.com/SomeoneElseOSM/SomeoneElse-style/blob/master/style.lua#L115

I personally wouldn't use OpenStreetMap's "standard style" as a hiking map. We have in the past had complaints such as https://help.openstreetmap.org/questions/20339/gordale-scar-malham-yorkshire-footpath . The mapping of http://www.openstreetmap.org/way/43323982 was actually excellent, but the standard style couldn't convey the likely problems to users.

(1) That's an England-and-Wales thing - don't worry about it elsewhere.

Potlatch editing 5 months ago

You've probably seen it already, but just in case not - should be fixed now:

https://lists.openstreetmap.org/pipermail/talk/2015-December/075245.html

As an aside, @Vincent%20de%20Phily - one of the reasons that I personally don't use JOSM for "normal" mapping is that I capture everything in a GPX trace (most information as waypoints), and it's difficult or impossible to do anything useful with that information in JOSM. There are a bunch of questions that I asked about this sort of thing ages ago - see the questions at https://help.openstreetmap.org/users/387/someoneelse that are unanswered (or have an answer of "you can't") and are tagged JOSM. It's possible that JOSM functionality has caught up since those were asked, of course. The main problem for me though is that JOSM's default UI just makes simple things (e.g. seeing object tags, viewing relations in the map view) very difficult or clumsy to do. I suspect that it might be able possible to play around with CSS to address some of the "what things look like" issues, but I've never needed to fix the problems as for me a better alternative already exists.

Sidewalks and crossings 5 months ago

@ksetdekov It depends on location, really. In many places in the world for pedestrian routing it makes no sense to map separate sidewalks and crossings because there are no explicit crossings - you're allowed to cross anywhere.

Where it gets complicated is where there are multiple sets of users, for example (1) pedestrians, who can and do cross anywhere (for whom sidewalk=none/left/right/both on the road is the best tagging) and (2) wheelchair and mobility scooter users who have to navigate by dropped kerbs, which may be at junctions or may not, and don't necessarily match any marked crossings that might exist. I've not seen a good solution that addresses both groups of users, but I do know (because I regularly use OSM data for pedestrian routing) that mapping sidewalks as separate ways, if done badly, can break pedestrian routing for everyone.

Mapping in the Japanese countryside 5 months ago

Hehe - finally a Mapbox diary entry where the animated gifs actually make sense!

In most cases animated gifs just make the entry unreadable (or, as happened recently, crash the browser), but here it's at least relevant :)