Wikipedia talk:WikiProject Highways/Archive 4

From Wikipedia, the free encyclopedia
Jump to: navigation, search
Archive 1 Archive 2 Archive 3 Archive 4 Archive 5 Archive 6 Archive 7

RFC on coordinates in highway articles

The following discussion is closed. Please do not modify it. Subsequent comments should be made in a new section. A summary of the conclusions reached follows.
The consensus of this RfC is section 9 to use shapefile software to illustrate the the area of highway mentioned in the article. Although I did not find that the canvassing to be detrimental to the RfC, I did find it disruptive to this RfC and do not support the use of such methods (since around 130 talkpages were canvassed in the sweep on the 17th). -- DQ (ʞlɐʇ) 18:39, 4 February 2012 (UTC)

Should coordinates be included in highway articles? If so, how should this be done, in terms of 1) what points of the road should be tagged or how certain roads are tagged and 2) the style that the coordinates should be presented in? 01:25, 26 December 2011 (UTC)

Proposals may be added by anyone, but should address the questions above. Please indicate Support, Oppose, or Neutral for each proposal that you wish; you may also indicate "first choice", "second choice", etc. Any relevant discussion is welcomed; any irrelevant discussion may be collapsed.

The actual wording of changes to WP:RJL or other guidelines / project standards will be determined at a later date; this will address the main principles.

This RFC will run for 30 days 31 days including the time the English Wikipedia is locked due to the SOPA initiative, at which time the RFC bot will remove the RFC template. At this point, a post will be made at AN(I) requesting closure by an uninvolved admin.

Please make sure that any crossposting to users, WikiProjects, and project noticeboards remains neutral and in compliance with WP:CANVASS. If you need help in doing this, there are some good instructions at WP:RFC, or you can use {{please see}}. Any violations of this will be noted to the closing administrator. --Rschen7754 01:25, 26 December 2011 (UTC)

Note: The RFC will be closed after 01:25, 26 January 2012. --Rschen7754 03:10, 17 January 2012 (UTC)

Proposal 1: No coordinates - Original research

Proposal 2: limited use of coordinates

Proposal 3: Start and end points of highway only

Proposal 4: Route Table

Proposal 5: Status Quo

Proposal 6: Strengthen use of coordinates in MoS

Proposal 7: Minimalistic entry in existing junction table

Proposal 8: (Live and let live)

Proposal 9: Shapefiles

Maybe we're going about this the wrong way. What would everyone think about attaching a polyline shapefile to articles that could be downloaded and used in any software that supports them? This would support at least 70 million coordinates per feature, not lead to any clutter beyond one link, faithfully represent the feature, and major features would able to be labeled via the attribute table. The shapefile specification is well known so it would be supported by many tools, and it would be trivial for editors to rip the coordinate data from already freely available shapefiles (in fact, the same shapefiles we use to create the maps in the articles!). It is less parsable than a single set of coordinates, but then with as many coordinates as would be needed to accurately represent the course of a major road, you would run into parsing problems with tons of {{coord}}s anyway. A microformat could be used to tag the shapefile link as containing geographical data so it would be machine-discoverable. Humans, too, could import the shapefile to their own GIS projects and render them against satellite imagery or in any projection we choose; our data could be used to augment other free maps. (Note! See the reply to comment #2 below for a possible way that Geohack could handle this.) —Scott5114 [EXACT CHANGE ONLY] 05:51, 29 December 2011 (UTC)

Discussion of Proposal 9

Proposal 10: Coordinate clutter

Proposal 11: Start & End coords in infoboxes are always welcome

Overall discussion

Request

possible SVG proposal

Wikipedia:SOPA initiative

Potential canvassing issue

Continuation of discussion

Template:shc

This RFC

Extending the RFC

Coordinates and OR

Surprisingly, perhaps, the proponent's enquiries elsewhere on coordinates and OR have not been reported by him here. See:

Is using a GPS receiver to find coordinates original research? and Coordinates and original research. While the original questions were framed in terms of GPS, the answers also touch upon the use of coordinates obtained from maps. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 00:07, 26 January 2012 (UTC)

There are still over 20+ editors that disagree, listed above. Consensus can change. - ʄɭoʏɗiaɲ τ ¢ 13:30, 26 January 2012 (UTC)

Closure of RFC

I've put in a request to get this RFC closed at WP:AN/RFC; unfortunately, it seems that we're at the bottom of a backlog, so it could be a few days. --Rschen7754 07:14, 26 January 2012 (UTC)

Ok. Does that still mean that the cut off time was 01:25, 26 January 2012? DubhEire (talk) 11:36, 26 January 2012 (UTC)
I think people can keep posting until the RFC closes. --Rschen7754 18:14, 26 January 2012 (UTC)

Another reason to include coordinates

Wikipedia now has a mobile app (for Android and iPhone). v1.1, currently in alpha for Android, allows users to tap any instance of {{Coord}}, and then be shown a map with five pins, linking to, respectively, the five nearest geotagged articles. This allows, for example, a reader planning a trip, to view the article on the highway they will be travelling on, select the coordinates for the junction at which they will leave, and then see what other points of interest are in the vicinity. Even without this feature, coordinates on articles are being used by version 1.0 around 27 thousand times a day, since its launch last month. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:51, 2 February 2012 (UTC)


The above discussion is closed. Please do not modify it. Subsequent comments should be made in a new section.

Moving forward on Proposal 9

Since Proposal 9 is another proposal with near unanimous support, I think we should likewise take a look at how to move forward with making it happen in the real world. Here are some of the questions that I think need to be agreed on:

  1. What sort of format are we looking at for the coordinates? Proposal 9 itself mentioned "shapefiles", but KML seems to do the same job, and much easier, too. Does anyone really object to using KML over actual shapefiles?
  2. Assuming KML is used, is there any way we could place the KML on a wikipage, perhaps a subpage of the talk page? Would this present unwarranted parsing difficulties?
  3. If we cannot put the KML on a wiki page, who do we need to convince to allow us to upload KMLs/shapefiles to Commons or the English Wikipedia?
  4. How will the link to the coordinate file be presented in the article? Will it be visible in the rendered output? Can it be tagged in some consistent way so that reusers of data can consistently find it? (Perhaps a class or id parameter in the HTML, somewhat akin to the way microformats are specified?) Would it be feasible to allow the file to be named anything (i.e., must it always be named <article title>.kml)?
  5. How can this be tied into GeoHack? How will the GeoHack link, if any, be presented in the article?
  6. Likewise, how can this be tied into the WikiMiniAtlas? How can we use the WikiMiniAtlas to enhance our articles?
  7. Who do we need to talk to in order to get the ball rolling on the necessary coding?
  8. Do the technical people need anything from us? Can we help them in any way to make implementation as smooth as possible on their end?

I think that about does it. If there are any questions that need to be addressed besides those, now is the time to get them ironed out. —Scott5114 [EXACT CHANGE ONLY] 21:05, 26 January 2012 (UTC)

The more I'm looking into it The more I'm getting convinced that the data duplication that results from uploading shape files (not mentioning issues with proper licensing) are a bad thing (tm). I suggest focussing on a way to refer to OpenStreetMap ways. This is not completely trivial, but reduces redundant efforts. --Dschwen 23:29, 26 January 2012 (UTC)
Could the resultant file be dynamic from data in hidden or not on the page/template? Could it even be stored via the toolserver rather than commons thus reducing the effort required to maintain? I know you probably can't answer straight away. Do you have an example of a KML (for example) that has a route following a road and some points on it. Must find/make one for curiosity. DubhEire (talk) 00:03, 27 January 2012 (UTC)
Sorry, but wouldn't it be fair to say that the complexity here is the route information, i.e. not all roads are perfectly straight. Bringing in other points of interest would make it more of a challenge also. DubhEire (talk) 00:27, 27 January 2012 (UTC)
I don't think licensing is really an issue here. The roads projects already use shapefiles for creating raster maps; we are always careful to use free data sources. Freely licensed shapefiles are not all that hard to find; many state governments release their stuff as PD, as do some universities. What data is being duplicated? As for OSM ways, I don't really think that is necessarily a good idea, considering ways do not necessarily correspond to any given numbered route, and I'd hate to have to impose on them by saying "we need this way split over on Wikipedia even if you don't have any reason to split it". Furthermore, I personally would like to avoid getting entangled in OSM any more than I have to; I don't understand their community or how their tagging is supposed to work and it seems really inconsistent (i.e. the suburb I live in is tagged completely differently than the main city 20 miles away, the same road classifications mean different things between the two cities, etc). I don't think it is really reasonable to ask editors here to learn an entirely new set of cultural norms and editing procedures if they should want to add geographic data to their articles. —Scott5114 [EXACT CHANGE ONLY] 15:19, 27 January 2012 (UTC)
A shapefile is a standardized presentation of GIS data, which is measured by surveying. They can't be copyrighted, as there is no presentation, just measurements. - ʄɭoʏɗiaɲ τ ¢ 15:36, 27 January 2012 (UTC)
You should be careful with statements like that. Databases enjoy special protection. --Dschwen 15:43, 27 January 2012 (UTC)
Feist v. Rural: the raw data can not be protected by copyright in the US. Imzadi 1979  15:45, 27 January 2012 (UTC)
Scott, how OSM tags their ways is not an issue to us. We would not have to meddle with their tagging or their splitting of ways. It just boils down to formulating a query that obtains a certain set of nodes and lines. Such a query could do the splitting as well. We'd get raw data, not their road classifications. And such a query would still be smaller than painting your own road or copying it. It could fit into a template. --Dschwen 15:49, 27 January 2012 (UTC)
How would this work in practice? Say, if I need only half of a way, because the route I am interested in leaves midway through a way? —Scott5114 [EXACT CHANGE ONLY] 22:01, 27 January 2012 (UTC)
Give me an example for a problematic route and I will build you a query. --Dschwen 20:12, 28 January 2012 (UTC)
Sooo... this discussion is already dead again? I guess the candle that burns twice as bright burns twice as fast... --Dschwen 16:55, 1 February 2012 (UTC)
Like it. Point well made. I've been busy as I am sure you have too. I would still like to make an example so let's define what that should be in abscence of anything else. Are you looking to build a query on the toolserver to do this? I was thinking of manufacturing one by hand and then layer in anything else. Do you have an idea youself what steps must be taken? I am not that familiar with how things work on the toolserver, but I am willing to try. DubhEire (talk) 17:03, 1 February 2012 (UTC)
There is query to map (which I worked on as well a long time ago). It is abit outdated, so i might have to create a new version or construct an SQL-query on the database by hand. But this should get us started. All I need are examples for problematic ways. You must understand that my knowledge of the highway business is limited, so I may have an oversimplified idea of what is necessary to get highways out of the database in a manner that will please the pros. --Dschwen 17:22, 1 February 2012 (UTC)
I see how it works. It's too bad that, for the most part, the routes I care about have not been plotted on OSM. –Fredddie 18:46, 1 February 2012 (UTC)
Really?! What kind of routes are that? --Dschwen 22:44, 1 February 2012 (UTC)
I should clarify. The routes I care about don't have an OSM relation, which are made up of ways. A few roads articles have OSM relations tagged using {{Osmrelation}}. That's what I meant, sorry. –Fredddie 23:25, 1 February 2012 (UTC)
But you don't need osm relations, tags on the ways can be sufficient to query the data. --Dschwen 05:03, 2 February 2012 (UTC)
OK, I guess I need a picture drawn. Let's use Iowa Highway 415 as an example, since I was looking at it in OSM earlier today. It's shown on the map as IA 415. As a bonus, it's not a simple straight-line highway. –Fredddie 05:11, 2 February 2012 (UTC)
Good, I'll look into that. I'm moving today and won't have reliable internet access for a few days. But I'll be back! --Dschwen 13:25, 2 February 2012 (UTC)

──────────────────────────────────────────────────────────────────────────────────────────────────── Visualization pending (this is all I can quickly do from a hotel lobby:

osm_mapnik=> select name from planet_roads where ref='IA 415';
          name            
---------------------------
West Bridge Road
Northwest Polk City Drive
Northwest Polk City Drive
Northwest 35th Street
Northwest 35th Street
Northwest 35th Street
Northwest 78th Avenue
Northwest 78th Avenue
Northwest 78th Avenue
Northwest 78th Avenue
Southwest Oralabor Road
Southwest Oralabor Road
Southwest Oralabor Road
Southwest Oralabor Road
Southwest Oralabor Road
Southwest State Street
Southwest State Street
Northwest 2nd Street
Northwest 2nd Street
Northwest 2nd Street
Northwest 2nd Street
Northwest 2nd Street
Northwest 2nd Street
Northwest 2nd Street
Northwest 2nd Street
Northwest 2nd Street
Northwest 112th Street
State Highway 415
State Highway 415
State Highway 415
Mile Long Bridge
(31 rows)

Does this look right to you? --Dschwen 02:43, 3 February 2012 (UTC)

After I realized it wasn't a serial list of ways, yes it does. –Fredddie 04:24, 3 February 2012 (UTC)
A couple questions:
  • I notice a "where ref='IA 415'" on there. Is this an attribute on OSM that we are using to snag just a part of a way or just a variable to aid in processing on our end? If it is the former, how would we proceed if this tag is missing on OSM?
  • What will happen if the data is changed on OSM? Will we have to alter the query on our end too?

The missing data question is most important to me, as there are situations that we cover on Wikipedia that OSM would not have tagged; unsigned highways, for example. An instance of this is Interstate 444, which is a snippet of freeway in downtown Tulsa. Because of all the highway designations that converge here, I-444 signs are left off the road, but it is officially on the Interstate rolls and therefore we cover it with its own article. However, since knowing the road is I-444 is useless for actual navigation on the ground, OSM does not have any sort of I-444 tags at all; they mark it the same way it's marked on the ground, as portions of US-412 and US-75. Another case I am concerned about is rural areas where perhaps the data has not been tagged or processed at all since initial import from TIGER (i.e. it is simply tagged as whatever the Census Bureau tagged it with) due to lack of interest in that area by OSM editors (it happens on this wiki, I'm sure it happens there too). —Scott5114 [EXACT CHANGE ONLY] 18:43, 3 February 2012 (UTC)

ref is an osm attribute. And a way is really a way-segment. An OSM way is always a uniform piece of road, as I understand it. So there really never should be a case of needing to get part of a way. Of course we can kill this discussion with a bunch of unlikely ifs. Can te data change on OSM? Yes, but I would cache the query results, and cache clearing could be made in a way that the nre result is previewed and checked first, so that the query could be adapted. Might there be some exotic roads where the tags are still missing? Maybe, but I'd rather cover 95% of all use cases with a clean and rather simple solution, than continue discussing about a suboptimal solution. A suggestion: get an OSM account and add the tags onto the data yourself, you'll be doing the world a service at the same time ;-). Anyhow, I will need a few days to settle in at my new place, but I already played around with querying the actual coordinate data and plotting it in the browser (this is a lot of fun!). --Dschwen 14:24, 4 February 2012 (UTC)
Ways are always uniform pieces of road, but a route doesn't always follow them. Consider the very common and not at all exotic instance of a concurrency. This is a segment of road with two different designations following it, i.e. both Interstate 44 and State Highway 3. This will, in my experience, usually be tagged in OSM as just the more important designation, in this case, I-44 (makes sense for OSM's users, it's what most people will be interested in). In any event, if you were trying to get a route trace for OK-3, you're going to miss out on those extensive concurrent sections (the first 304 miles of OK-3 is concurrent with another road, and it's only independent for 40 miles or so before it joins up with something else) because they won't be tagged as SH-3, they'll be tagged as whatever the more important road is.
As I said before, though, I consider it unreasonable to have to require editors to have to edit data over there, learn their conventions for tagging, etc. Personally I have enough going on here on Wikipedia to have to worry about falling down a whole new rabbit hole of butting heads with OSM editors for reasons that do not matter to them. (Why should they care if I need so and so tagged in this or that way for Wikipedia's queries? It doesn't benefit OSM any.) I, and probably most of the other roads editors, would much rather prefer to be able to hand tune the data to meet Wikipedia's needs without having to step on someone else's grass. Before we spend too much time chasing the OSM thing, can we stop and investigate the KML option more thoroughly as well? I don't really see why we've abandoned the idea of uploaded KML files other than a vague "data duplication" explanation (which I don't think holds much water, frankly; what we'd be doing with the data is somewhat different than what OSM is doing—it's a subtle difference, yes, but a difference).—Scott5114 [EXACT CHANGE ONLY] 15:00, 4 February 2012 (UTC)
Feel free to chase the KML dream. I won't actively push for it. It makes no sense to me to reject the notion of collaborating with OSM to improve the tagging for everyone's benefit. I will look into the road concurrency thing, I don't see why OSM would not be interested in properly tagging those ways. It actually does not make sense for OSM users to tag only with one designation. Just think about routing applications. Saying those correction would be another rabbit hole and you wouldn't want to force users to edit on another project seems a bit absurd, considering the alternative of using complex external software to prepare the data. --Dschwen 15:55, 4 February 2012 (UTC)
You mean complex software like Google Earth, where you make a path and export to kml? - ʄɭoʏɗiaɲ τ ¢ 16:28, 4 February 2012 (UTC)
That will most certainly give you a ridiculously low quality file, unless you spend a lot of time... ...duplicating the efforts that were already made elsewhere. Not invented here syndrome. --Dschwen 17:05, 4 February 2012 (UTC)
Dschwen, we already use GIS software and high-quality, freely-licensed GIS data in the creation of our maps. It's not terribly complex for us to select the red line that has just been created for the purpose of a map and export it to a KML. We teach new editors that are interested in the mapping aspects of the project how to use QGIS whenever they ask. We have tutorials written. We have the infrastructure for this ready on our end; don't worry about the complexity there. —Scott5114 [EXACT CHANGE ONLY] 17:18, 4 February 2012 (UTC)
Road concurrency is taken care of at the highway level at least in OSM (see I 57; I 70 near Effingham, IL). However this turns out, I'll be happy to add overlay support to the WikiMiniAtlas. --Dschwen 17:06, 4 February 2012 (UTC)
I was trying to use the Query to Map method of retrieving OSM data for river liffey or N11 in Ireland. I'm not sure if I am querying it correctly. I tried Way:name=Liffey ;Node:river=* What do I put the bounding box as? I tried -5,52.-6,53 . I also tried Way:name=N11 ;Node:highway . Also tried leaving bits out too. Just trying to understand this approach a bit better. Thanks. DubhEire (talk) 11:24, 18 February 2012 (UTC)
That particular tool is just broken. Above I gave an example for how to manually query the database. However the OSM querying approach was shot down, so I did not pursue that avenue any further. --Dschwen 02:26, 19 February 2012 (UTC)