[RFC] Feature Proposal - landcover proposal V2

A proper solution could be to store references to tags instead of tags themselves.

That way when a tag change is proposed all older tags will be updated in the new schema but still work in the old one, and all new tags will be backwards compatible with the old schema.

So, proposal authors could specify with tag a new tag will be in the old schema. For example, you could specify that highway=busway would look like highway=service for data users on the old schema.

2 Likes

I rather interpret this conversation as the landcover proposal not being viable within the current context (i.e. the absence of a deprecation/versioning process). That is, it makes sense to put it on ice for now and instead focus on versioning etc. (possibly using it as a usecase), but then re-visiting it when we have the infrastructure (both technical and non-technical) to handle such a change.

1 Like

can you illustrate with landuse=farm how this would help solve the issue?

1 Like

While a tagging scheme such as one of these is possible, “versioning” can mean many things (and I think even calling it “versioning” might be limiting, as there might be approaches that don’t really have anything to do with versioning but that solve the same problem).

Earlier in the thread I posted a few possible approaches:

  • Providing planet dumps that are “normalized” according to a specific version, a new version would come every X years, and dumps for the last Y versions are available
  • Provide “normalization” scripts in various languages so that each consumer can choose their target language which the rest of their system works with, without having to manually see what has changed over time
  • Just let it play out, but disallow mass edits, so that consumers would start to see gradual changes over time, and hopefully adapt on their side
  • Having official “version change days”, one every X years, on which all data is adjusted according to the new tagging scheme, and made very clear to all consumers that this will occur and on which days

All these, in various ways, solve the same problem (migrating between tagging schemes), all have trade-offs, and all will have (vocal) opponents. As @stevea says, devising this is a huge undertaking (but an important one, possibly even one of the most important ones OSM will undertake if it wants to stay relevant).

2 Likes

“The current context” is OSM right now. The Proposal is effectively “dead on arrival” (sorry to be harsh in my wording) with the structure (both social and technical) OSM has right now. It would certainly need to be updated (as a v3?) if and when a comprehensive (clearly longer-term) solution might emerge along the lines you suggest. But to put it “on ice” as it is now and expect that we could simply “thaw” it with what effectively would be a new paradigm for OSM tagging is unrealistic.

Just my opinion.

:popcorn: :popcorn: :popcorn:

(since I can’t use it as a reaction…)

It was never viable. Not because the discussion here is long, but because the proposal lacks a fundamental understanding of the technical landscape of the project. The length of the discussion simply serves as a body of evidence for the next time someone proposes an unreasonable change.

2 Likes

That’s pretty much exactly what I said :stuck_out_tongue: It is not viable right now, but it (or rather a future iteration of it) might be once OSM “is ready for it”.

It might be possible to just pick up where it is now, it might need to restart from scratch (that will depend on how a tagging-migration-solution ends up being), but regardless one should look at what was already done and discussed.

1 Like

Then, solve it, and be prepared to show your work. Whether this is a dismissal or a call to action only depends on your personal level of interest and motivation to roll up your sleeves and demonstrate a solution. I’ve certainly been told that “if you don’t like it, do it better” and then gone off and done so on this project. I’ve built software, executed hard tagging changes, and spoken at conferences and on YouTube on my exploits. And, I’m nobody. Just a hobbyist that felt like it.

Your language is in reverse, I think. OSM is not some entity which you sit outside of and chastise it to do better. OSM is you. It’s me, it’s @stevea, it’s @SomeoneElse entirely. It’s a software developer, an end user, it’s the OSM Foundation Board, a working group, a local chapter, an activist, a hobbyist. It’s nobody in particular. But it IS individuals that have seen problems and said “I can do better” and come up with the goods. That’s the reality in a decentralized project.

“OSM” doesn’t undertake anything. Individual contributors do.

1 Like

true, but is them up to them to find and fix the issue before distributing updated dataset to their users.
IOW, (theoretical) improvement of using such dataset would be that when tagging style changes, only one place (Overture or whoever) has to worry about it and do backwards-compatible fix; instead of every data consumer having to handle it.

Of course, it also creates another set of problems, not least of which being that tagging structure is ossified there, and you can not do any improvements in data tagging, because if you change or introduce any new type of data, you’re back on square one. Also since one had to check if there has been vandalism or new or changed tags before releasing, there would be delays in updated maps, but that can be an advantage too sometimes.

An actual versioning and backwards compatibility solution for OSM would allow a period of time where mappers could submit either v1 or v2 tags to the API and consumers could get data out with either v1 or v2 tags. A translation layer on both ends would be required. With the tag agnostic philosophy of the project this seems unlikely to happen.

If someone would like do such translation (which I don’t see as an idea with particularly useful); may I recommend (instead of such ugly extra-tag-prefix/suffix kludge) to instead create new API endpoint instead.

e.g. currently this API endpoint https://www.openstreetmap.org/api/0.6/relation/5876272 returns some XML. One could easily create implement translator API at https://whatever.example.com/api/0.7/relation/5876272 which would return that object in the way that they find preferred (e.g. replacing all natural=wood or landuse=forest to landcover=trees or whatever other change and reasoning one might have to deprecate/rename tags), e.g. it would “translate” regular OSM-ATYL-tags to my-more-rigid-and-clearer-set-of-tags, like this:

<?xml version="1.0" encoding="UTF-8"?>
-<osm version="0.6" generator="CGImap 0.8.10 (412697 spike-07.openstreetmap.org)" copyright="OpenStreetMap and contributors" attribution="http://www.openstreetmap.org/copyright" license="http://opendatacommons.org/licenses/odbl/1-0/">
+<osm version="0.7" generator="My New Cool API 0.9.37  (116321 whatever.example.com)" upstream="CGImap 0.8.10 (412697 spike-07.openstreetmap.org)" copyright="OpenStreetMap and contributors" attribution="http://www.openstreetmap.org/copyright" license="http://opendatacommons.org/licenses/odbl/1-0/">
 <relation id="5876272" visible="true" version="6" changeset="118694112" timestamp="2022-03-20T12:24:14Z" user="Muki Mačković" uid="3636444">
  <member type="way" ref="87194173" role="outer"/>
  <member type="way" ref="392034935" role="outer"/>
  <member type="way" ref="753330854" role="inner"/>
  <member type="way" ref="1041941121" role="inner"/>
-  <tag k="landuse" v="forest"/>
+  <tag k="landcover" v="trees"/>
  <tag k="name" v="Park šuma Grmoščica"/>
  <tag k="type" v="multipolygon"/>
  <tag k="wikidata" v="Q97353235"/>
  <tag k="wikipedia" v="hr:Park-šuma Grmoščica"/>
 </relation>
</osm>

Then people can then easily change their JOSM (or whatever) configuration to use that new API endpoint; and enjoy in their new&improved OSM tagged worldview.
The same modifications (if wanted) can be made for objects being PUT on the server, if you prefer to upload only the “new and better” tags instead of the equivalent-tags that user has specified (you probably should create an explicit changeset tag that indicate you did such auto-translation, though)

The best thing? One can do that right now with minimal effort (certainly one or more orders of magnitude less effort than this discussion has entailed so far), and one can test and develop such project on any small server/VPS, and gather smaller (or bigger) number of enthusiastic beta-testers to improve and test it before proposing it as an official one. And even if it is not accepted as official one, who cares? The ones who elected to use it would still reap all the benefits that such tag deprecation/renaming (supposedly) entails.

3 Likes

I think you should look at the context of both times I posted those suggestions; the first time to exemplify that it is (or might be) possible to have a scheme that allows changing tags in a consumer-friendly way, the second time to exemplify other alternatives to prefix/suffix versioning.

I’d love to take a stab at this in a more serious manner, but considering that I already have two rather major OSM-related projects I’m working on right now which are already consuming most of my free time I have to politely decline.

I was referring to OSM as an “organism” or a community, as this is not something any one single person can do. Sure, it will require one or a few heroes who pull the community together around the task and do a large share of the work, but in the end it will go nowhere without the involvement of a large part of the community.

Do note that your comment comes across as somewhat condescending. I’m well aware what drives a project such as OSM forward.

1 Like

“landcover vs landuse” has been debated since at least 2010. This particular thread isn’t even a year old yet. :smiley:

OSM (with its completely free tagging) has succeeded where other competitors have failed, just like wkipedia succeeded after nupedia failed.

What’s allowed OSM to be successful is allowing anyone to describe their local area so that they, or other people, can make a map from it. Restrict that, and you restrict what people can map - which wouldn’t be a good thing. That doesn’t mean that there can’t be a role for a set of data with a more or less fixed schema so that the problem Brian described above is less of an issue, and yes, the schemas for OpenMapTiles et al do intend to nail things down a bit so that a consumer of (say) vector tiles claiming to be using OpenMapTiles’ schema don’t have “lots of data suddenly missing” - but what Zeke said above is true - and it does mean that if people suddenly have the urge to change the tag in OSM for concept X from tag A to tag B, they’d better have a pretty good reason. One to one changes that literally add no value are NOT a good reason.

Maybe an example would help. I look after a couple of different sorts of maps - web based ones (click the changelog link to see more details and links to the sources of the components) and Garmin / mkgmap ones, source here. Where possible both use more or less the same schema, so that the way that I round up all possible taggings of (say) “broadleaved woodland” is the same in both cases, but the resulting map style code is very different…

If someone decides to change all tag A to tag B in the data, each map will just “see fewer examples of broadleaved woodland”. I’ve seen that happen with embassies, hill forts and pinfolds. That’s despite both those map styles being registered at taginfo, with a link to both github projects so that people can create an issue saying “I’m proposing to change tag A to tag B” alongside anything else that might apply.

3 Likes

Oh, dear Andy our Support moderator, thank you for both your reminder of the longevity of this issue (and related ones about tag deprecations) — I remember 2010 in OSM, too — as well as your helpful personal experiences.

Taginfo “registrations,” GitHub project links that foster good communication and many other earnest, genuine attempts at “reconciliation” (for lack of a better word, though “harmonization” also gets close…I know you might prefer the z be an s) and more are all important ingredients at how this continues to move forward.

Though, for now, may we split this topic into a new one that is more aligned with tag deprecation and let this topic continue to discover (and accept) that the Feature Proposal that it purports to be about simply isn’t viable? Thank you.

1 Like

I’d love to see a feature proposal that genuinely added value - how about an area data type? :smiley:

You are thinking along the right lines, but retooling the API endpoint alone is insufficient. You would need to consider all the pipelines through which data consumers get their data. At a minimum you’d need to distribute parallel planet files and diffs, and likely stand up parallel overpass services to cover the most common pipelines.

All of this is assuming that the problem we identify is worth all this effort :slight_smile:

1 Like

You mean a primitive? I am aware of node, way, relation. Out of the relation type, multipolygon as far as I can tell only ever maps areas.

I think mapping intent is missing from the list.

8 posts were merged into an existing topic: Suggesting area data type

Where is that :popcorn: emoji reaction when you need it… Anyway, might I suggest splitting that area data type part of the discussion to another subthread?

Sure, but only because (as you said) it has neither area=yes nor area=no tag. If you know of the real situation, you could just add area=* tag, and the problem would be solved, right? Or am I missing something else there?

1 Like

And it could be done, in my opinion, without an API change. You could have some code between the API and the database that rejects changes that turn a valid multipolygon into an invalid one. Because really, we have areas, they’re just multipolygon relations that have no integrity guarantee. The difference between not allowing users to break multipolygon relations and actual area types in the API is, in my opnion, just syntactic sugar around the core problem of ensuring data integrity.

Of course, like this proposal for landcover, areas aren’t that interesting because they doesn’t solve a significant problem. After all, data consumers are handling areas just fine – though a little extra data integrity would be nice.

Actually, can we rename this thread to “idealistic ideas to improve OSM” ?

:popcorn:

We could try and “make a rule” that all areas need to be mapped as relations (not ways), but I suspect that there would be considerable resistance to that change…

(for the avoidance of any doubt I would not be in favour of any such rule)

Can someone then create another topic with “idealistic ideas to improve OSM” or should I?
So as not to divert the main topic here.