Should iD be tagging `crossing=uncontrolled` instead of `crossing=marked`?

[joepie91]

Currently, the 'marked crossing' preset will tag the crossing with crossing=marked. The wiki page for crossing claims that this is a duplicate of crossing=uncontrolled, and that the latter is 'approved', and indeed in terms of usage statistics it appears to have seniority. The highway:crossing wiki page claims that both are semantically identical.

On the other hand, the wiki page for crossing=uncontrolled claims that marked vs. unmarked "seems to be more specific", though it fails to provide a clear rationale for that claim. It implies that uncontrolled is used for both marked and unmarked crossings currently, but this is not corroborated by other wiki pages on the topic, as far as I can tell.

Should iD be adopting crossing=uncontrolled as the canonical tag instead, that being the 'senior' tag which was already in wide use prior to iD's preset? Or is there a specific reason to continue using crossing=marked instead?

[Dimitar5555]

crossing=uncontrolled is more useful for railway=crossing. Also crossing=marked and crossing=unmarked provide more information compared to crossing=uncontrolled. In my opinion the use of crossing=uncontrolled on highway=crossing should be discouraged.

[joepie91]

crossing=uncontrolled is more useful for railway=crossing.

That doesn't sound right to me - it is primarily used for road crossings in practice, with railway usage being much less common, and was indeed approved for this use through a vote. What specifically makes it "more useful for railway crossings"?

Also crossing=marked and crossing=unmarked provide more information compared to crossing=uncontrolled.

According to the majority of documentation I've been able to find, the typical set of choices is traffic_signals (marked and signalized), uncontrolled (only marked), unmarked (not even marked), and no (explicitly forbidden). While a bit of a strange set of terms, this is the widely accepted use on OSM, and I don't see what marked adds to this - it would be essentially equivalent to uncontrolled, and we already have unmarked either way. (I do think the documentation could be clearer on this.)

If you feel that there is still some kind of additional context provided by marked that is not present in uncontrolled for some reason, could you elaborate on what that context is exactly?

In my opinion the use of crossing=uncontrolled on highway=crossing should be discouraged.

Is that not something that should be discussed on a community-wide level, considering the extremely wide adoption, rather than encoded into the presets of a major editor?

[1ec5]

crossing=uncontrolled and crossing=marked represent two orthogonal tagging schemes:

• The crossing=uncontrolled/traffic_signals scheme originally had no way to express an unmarked, unsignalized crossing. crossing=unmarked was later popularized as a workaround, but crossing=traffic_signals still assumes that any signalized crossing is marked, so the signalized, unmarked crossings common in some regions cannot be expressed.
• The crossing=marked/unmarked scheme relies on a separate crossing:signals key to indicate signalization. The crossing and crossing:signals keys together can express all four configurations.

iD supports neither scheme completely. Since 2019, it has had presets for crossing=marked and crossing=unmarked but no field for crossing:signals. In #192, a preset was added for crossing=traffic_signals but not for crossing=uncontrolled.

For better or worse, crossing=marked/unmarked has achieved “de facto” status and outpaced the “de jure” crossing=uncontrolled/traffic_signals by a wide margin. The 2008 vote you cited wasn’t a particularly decisive vote. There were almost as many complaints about UK-centrism and inconsistency with existing usage as there were votes in favor, and the outcome explicitly sought a more universal tagging scheme. The proposal also focused on pairing crossing=uncontrolled with crossing_ref=zebra. (The “zebra” terminology was removed from iD in 2019 after much controversy.)

crossing=uncontrolled is one of OSM’s more prominent misnomers. In English, an uncontrolled crossing means an unsignalized, unmarked crossing, whereas crossing=uncontrolled is only for unsignalized, marked crossings. Data consumers can only interpret this tag as meaning unsignalized, without any confidence in whether it’s marked or unmarked, because mappers may very likely have used either definition.

If iD were to simply add a preset for crossing=uncontrolled alongside the existing ones, even an unsearchable one, mappers would still need to intuit the differences between the various crossing presets. “Uncontrolled Crossing” would be misleading as described above. “Marked Crossing Without Pedestrian Signals” would be quite wordy, “Marked Crossing Without Pedestrian Signals (Raised)” even moreso. One is led to wonder why the “With/Without Pedestrian Signals” presets shouldn’t be relegated to a field indicating whether there’s a pedestrian signal at the crossing.

[Dimitar5555]

That doesn't sound right to me - it is primarily used for road crossings in practice, with railway usage being much less common, and was indeed approved for this use through a vote.

There are far more road crossings than railway crossings, hence why it has more uses.

What specifically makes it "more useful for railway crossings"?

I don't know about you but I prefer to not play Russian roulette while crossing railway tracks.

If you feel that there is still some kind of additional context provided by marked that is not present in uncontrolled for some reason, could you elaborate on what that context is exactly?

marked implies that the crossing has zebra crossing which gives priority to the pedestrians (it may not apply in every country), while on unmarked crossings pedestrians are required to wait for cars to pass (may vary from country to country).

[joepie91]

• The crossing=marked/unmarked scheme relies on a separate crossing:signals key to indicate signalization. The crossing and crossing:signals keys together can express all four configurations.

Aaahhhh, that is the missing piece of the puzzle then. In trying to figure out crossing tagging conventions, I haven't run across anything that made this clear - I was under the impression that the new tagging scheme was "traffic_signals vs. marked vs. unmarked", rather than "marked vs. unmarked with a separate axis for traffic signals entirely".

For better or worse, crossing=marked/unmarked has achieved “de facto” status and outpaced the “de jure” crossing=uncontrolled/traffic_signals by a wide margin.

Is that not specifically because of iD adoption, though? The impression I've gotten is that the rest of the community is mainly sticking to uncontrolled rather than marked, and indeed uncontrolled usage is also still growing, albeit not as fast.

crossing=uncontrolled is one of OSM’s more prominent misnomers.

I had not run across this proposal before, and it brings up a number of good points. I'm curious why it seems to have gotten stuck.

marked implies that the crossing has zebra crossing which gives priority to the pedestrians (it may not apply in every country), while on unmarked crossings pedestrians are required to wait for cars to pass (may vary from country to country).

This is not any different from or more informative than uncontrolled vs. unmarked, when used in the traffic_lights/uncontrolled/unmarked sense that the wiki suggests.

---

Overall: while I do not find all of the arguments presented convincing, I think there are enough good arguments to alleviate my original concern of semantic duplication, and that there's no need to immediately change anything in the iD presets. So I suppose that the original question in this issue has been answered.

That having been said:

• The documentation on the wiki is clearly insufficient. It's not at all clear that marked/unmarked defines traffic light presence on a separate axis entirely, and the current description of marked as being semantically equivalent to uncontrolled is misleading. This will probably require changing in several different places, if it is indeed true that traffic light presence is meant to be separate from the crossing type entirely. One of the linked proposals does seem to confirm this.
• I do think that there should be a real push to gain community consensus on this new tagging approach, deprecating the previously-approved one in the process. Things seem to be in a sort of strange not-quite-approved limbo right now, and that would make tag upgrading efforts difficult, and creates confusion for mappers in choosing a tagging scheme.
• iD should probably implement the traffic signal component of the new schema as well, then?

The first two points are out-of-scope for this repository, though.

[Dimitar5555]

The impression I've gotten is that the rest of the community is mainly sticking to uncontrolled rather than marked, and indeed uncontrolled usage is also still growing, albeit not as fast.

Most people tend to use the presets provided by their editor.

Is that not specifically because of iD adoption, though? The impression I've gotten is that the rest of the community is mainly sticking to uncontrolled rather than marked, and indeed uncontrolled usage is also still growing, albeit not as fast.

Replace "the rest of the community" with "JOSM users".

This is not any different from or more informative than uncontrolled vs. unmarked, when used in the traffic_lights/uncontrolled/unmarked sense that the wiki suggests.

traffic_lights/marked/unmarked is a lot clearer and unambiguous.

iD should probably implement the traffic signal component of the new schema as well, then?

But before that the proposal should pass through the full proposal process.

[davidpnewton]

To my mind it definitely needs orthogonal tagging to describe situations. In fact I'd argue there is more than one orthogonality needed.

Marked/unmarked to describe whether there are markings on the road surface defining the crossing.

Raised/not raised to describe whether the crossing is on top of some kind of associated traffic-calming bump or not. For example near where I live there are examples of Puffin crossings (traffic signal controlled marked pedestrian crossings) and unmarked segregated cycle track/pedestrian crossings being associated with speed bumps.

Pedestrian only/cyclist only/pedestrian and cyclist in one space/pedestrian and cyclist in separate spaces. Again to use a UK example Toucan crossings are traffic signal controlled marked pedestrian and cyclist crossings where pedestrians and cyclists are not segregated on the crossing whereas parallel crossings are marked pedestrian and cyclist crossings where pedestrians and cyclists are segregated on the crossing.

Pedestrian/cyclist/pedestrian and cyclist to define who can use the crossing.

Controlled/uncontrolled is an unhelpful distinction nomenclature because it encompasses more than one characteristic in its "distinction" attempt. It is thus ambiguous and open to serious misinterpretation.

Re-posted to the correct issue this time!

[1ec5]

• The documentation on the wiki is clearly insufficient. It's not at all clear that marked/unmarked defines traffic light presence on a separate axis entirely, and the current description of marked as being semantically equivalent to uncontrolled is misleading. This will probably require changing in several different places, if it is indeed true that traffic light presence is meant to be separate from the crossing type entirely. One of the linked proposals does seem to confirm this.

Good point. I wrote a new overview article about the different kinds of crossings and the conflicting crossing classification schemes. I also explained the crossing=unclassified misnomer in more detail in the article for that tag. I tried not to get into too much detail about iD’s presets and the controversies surrounding them over the years, since that seemed to be covered at length on individual tag pages. My focus was on documenting the overall schemes conceptually. I hope this is will be informative to mappers and others looking to understand the state of affairs and hopefully a starting point for arriving at a consensus in the future.

[joepie91]

Good point. I wrote a new overview article about the different kinds of crossings and the conflicting crossing classification schemes. I also explained the crossing=unclassified misnomer in more detail in the article for that tag. I tried not to get into too much detail about iD’s presets and the controversies surrounding them over the years, since that seemed to be covered at length on individual tag pages. My focus was on documenting the overall schemes conceptually. I hope this is will be informative to mappers and others looking to understand the state of affairs and hopefully a starting point for arriving at a consensus in the future.

Thanks! That definitely helps. A few minor nitpicks that I ran into while researching crossings (originally intending to write an article about this myself):

• There seems to be a regional difference in whether to tag the node where a cycleway crosses a road at all. In the Netherlands, the general advice seems to be "don't do that, as it'll be rendered as a foot crossing, and bikes have no special priority in NL". Elsewhere, at least in the Anglosphere, the advice seems to be "definitely do mark the nodes as crossings".
• The table on that page currently equates marked to crossing_ref=zebra, but there are other types of marked crossings that are not zebras.

[1ec5]

Thanks for taking a look. Please help improve the page or make further suggestions on the article’s talk page so we can keep this repository on topic. I detailed three tagging schemes but there are other tags that don’t fit neatly into any of these schemes, especially given how crossing_ref is a freeform slot for making regional distinctions. For that matter, I skipped over the topic of road signalization at crossings, which to my knowledge has no solid answers but lots of approaches in practice.

[hungerburg]

Just to chime in, here a JOSM user in favour of "marked" vs. "uncontrolled", though not so much a nomenclatura suprematist to rule out such a nice value as "traffic_signals". On to the message:

I very much doubt, that there are in fact any "uncontrolled" railway crossings, at least not in the Western world, as all "valid" railway crossings are at least signalised, at the minimum by blowing the horn, that is, when coming near the place. If it that tagging really is widespread, I suspect, people use this for stuff, where actually "crossing=no" should be the legally compliant tag, plus "informal=yes" on the crossing path, the OSM way of mapping such.

[1ec5]

I very much doubt, that there are in fact any "uncontrolled" railway crossings, at least not in the Western world, as all "valid" railway crossings are at least signalised, at the minimum by blowing the horn, that is, when coming near the place. If it that tagging really is widespread, I suspect, people use this for stuff, where actually "crossing=no" should be the legally compliant tag, plus "informal=yes" on the crossing path, the OSM way of mapping such.

In urban areas of North America, a tram or light rail crossing for pedestrians can be subject to a whistle ban or quiet zone and may lack crossing gates and lights. The only form of traffic control is that light rail vehicles are restricted to 5 miles per hour along adjacent tracks. There’s a preset for railway=tram_crossing apparently because these situations are the norm in some regions. Private crossings of heavy rail tracks typically lack gates and lights, and trains may not routinely sound their horns at them either.

Outside of the Western world, uncontrolled railway crossings are common in Vietnam, for example, although they may not be reliably tagged as such. They aren’t quite informal, because the authorities put up level crossing warning signs when they discover these crossings.

Anyways, most of the crossing=uncontrolled usage is on street crossings, not railroad crossings. There’s a lot of confusion among U.S. mappers about the meaning of crossing=uncontrolled. This proposal doesn’t exaggerate the confusion at all. For example, of the nearly 5,000 crossing=uncontrolled nodes in Silicon Valley, I’d estimate that at least 10% are unmarked (example, example). It’s impossible to say how many of the marked crossings were tagged with the same understanding that an uncontrolled crossing can be either marked or unmarked. These are all legal crossings. In fact, state law says uncontrolled (unmarked, unsignalized) crossings are implied at many, many unsafe locations where local mappers would hesitate to add a crossing; mappers have instead focused on those that have curb cuts or signs.

[davidpnewton]

"I very much doubt, that there are in fact any "uncontrolled" railway crossings, at least not in the Western world, as all "valid" railway crossings are at least signalised, "

Erm nope. Totally wrong. There are hundreds of crossings in the UK alone which are completely uncontrolled. Now admittedly they tend to be foot crossings for the most part, but they are still completely uncontrolled. Completely uncontrolled as in walk along the footpath, open the gate at the near edge of the railway land, look both ways, cross the railway, open the gate at the far edge of the railway land and walk through it. There is such a thing as an uncontrolled crossing for roads in the UK as well however. This for example:

https://www.google.com/maps/@50.9681235,0.8970443,3a,60y,33.58h,85.08t/data=!3m6!1e1!3m4!1sAz5hThyp0ZDBNhBEmu4yVQ!2e0!7i13312!8i6656

[hungerburg]

If the train there does not give a whistle, then I stand corrected, and the example shows an uncontrolled crossing. Around my location, almost all railway crossings have been converted to underpass or overpass. On a recent visit to another part of the country, I heard train whistles, this kind of signalling still exists.

As to tram: Rail is all about signals, while tram operates on sight. They are very different.

PS: Picture of "controlled" railway crossing https://oekastatic.orf.at/mims/2019/23/95/crops/w=1280,q=70,r=1/176981_bigpicture_45195_img_7098_hell.jpg

[davidpnewton]

Correct about railway and tram when it comes to tram street running. However again you miss things:

https://www.flickr.com/photos/127699427@N02/30906536117

That's a railway station. It's also a tram. Trams on that section most certainly do not run using line of sight! There are also plenty of instances where trams utilise former heavy rail infrastructure which certainly opens up the possibility of running using signalling there as well. Not saying that they do use signalling on such sections, but that the possibility is open for them to do so.

[hungerburg]

One more note on the subject: I got told, that iD users mostly map from aerials, therefore the distinction "marked/unmarked" makes a lot of sense for them, as this is something, that can be very easily observed from aerials. A bit of wondering remains, how then are they spotting the unmarked ones? My guess, these come from connecting footways circling blocks, as I was told, that iD users prefer to map sidewalks as separate ways too.

crossing=uncontrolled and crossing=marked represent two orthogonal tagging schemes:

I do not understand the bit about unmarked yet signalised crossings. I myself map pedestrian infrastructure, not paint on a surface. So for me the sequence: unmarked < marked/uncontrolled < traffic_signals, shows a clear progression in the amount of money or effort that was spent on location and the (perceived) safety it offers to pedestrians. Maybe that is a third scheme? Whether to call a crossing marked or uncontrolled is of no concern in this scheme, one or the other is just a term to be learned by heart. Still, "uncontrolled" truly is not something that is easy to grab ;)

[Dimitar5555]

A bit of wondering remains, how then are they spotting the unmarked ones?

There are laws which define where pedestrians can cross. If there isn't a marked crossing at such place it's an unmarked crossing.

I do not understand the bit about unmarked yet signalized crossings.

There is no such thing as unmarked signalized crossings but there are signalized crossings with worn off paint.

[1ec5]

On a recent visit to another part of the country, I heard train whistles, this kind of signalling still exists.

Definitely, in fact, this proposal about whistle bans points out that whistles (or horns) are the norm at all public heavy rail crossings in North America unless specifically exempted.

One more note on the subject: I got told, that iD users mostly map from aerials, therefore the distinction "marked/unmarked" makes a lot of sense for them, as this is something, that can be very easily observed from aerials. A bit of wondering remains, how then are they spotting the unmarked ones? My guess, these come from connecting footways circling blocks, as I was told, that iD users prefer to map sidewalks as separate ways too.

Generally speaking, iD users use the same sources as JOSM users: aerial imagery, street-level imagery, surveys, local knowledge. Aside from sidewalk geometry, tactile paving on curb ramps make it easy to spot unmarked crossings. The curb ramps exist in reality whether the sidewalks are tagged on the roads or mapped as separate ways.

There is no such thing as unmarked signalized crossings but there are signalized crossings with worn off paint.

To reiterate, in the U.S. and possibly elsewhere, it’s common for a crossing to be unmarked and signalized by design. Sometimes this is because of a decidedly nonstandard treatment, like the unmarked crossings across these tiled intersections in Seattle. But completely standard intersections can also lack markings for crosswalks, such as this intersection in small-town Missouri.

_Note the black pedestrian traffic signal boxes mounted on the pole for the traffic lights, replete with explanatory signs and tactile paving, suggesting that these unmarked crossings otherwise conform to the latest national standards._

Another very common situation in the U.S. is that a signalized intersection has a pedestrian crossing that lacks its own pedestrian signals. Pedestrians must wait for the cars’ green signal before crossing or risk a fine for jaywalking. Sometimes these crossings are also unmarked, but the law still allows a crossing unless a sign prohibits it. For example, this is unmistakably an unmarked crossing based on wheelchair-accessible curb ramps that are visible in both aerial and street-level imagery. It may have originally been marked years ago, but the markings were later paved over and never replaced. There are no pedestrian crossing lights; instead, the crossing is controlled by the main traffic lights. It would be very dangerous to jaywalk here, at the end of a freeway off-ramp where cars routinely speed around the curve.

Curb ramps with reddish tactile paving; shadows cast by traffic signals for cars strung from a wire across the intersection.

I do not understand the bit about unmarked yet signalised crossings. I myself map pedestrian infrastructure, not paint on a surface. So for me the sequence: unmarked < marked/uncontrolled < traffic_signals, shows a clear progression in the amount of money or effort that was spent on location and the (perceived) safety it offers to pedestrians. Maybe that is a third scheme? Whether to call a crossing marked or uncontrolled is of no concern in this scheme, one or the other is just a term to be learned by heart. Still, "uncontrolled" truly is not something that is easy to grab ;)

Yes, I’m pretty sure “level of money or effort” was indeed the original concept behind the traffic_signals/uncontrolled/unmarked scheme. There’s a certain elegance to it, but it doesn’t account for regional and local differences in government funding priorities. Having personally used crossings of the sort described above, I can attest to them feeling less safe than marked ones, if only because motorists are less accommodating around them. But if mappers don’t classify crossings based on specific traffic control devices, then they’d be left to make subjective, tracktype-like judgments about level of control in many cases.

My examples are from the U.S. because that’s what I’m personally familiar with. But my point is that the messy state of crosswalk tagging is a consequence of early assumptions based on one region’s standards that have proven unreliable in a global context. It would be one thing if uncontrolled were merely a term of art, but in fact it’s a real term that’s been misused.

[hungerburg]

Thank you for the pictures, the crossing in small-town Missouri clearly shows, regarding pedestrian infrastructure, municipal administrations can get by with less: funding, effort and (perceived or actual) safety. Myself, from a country that is party to the Convention on Road Signs and Signals, I'd not consider such a feature up to the local standards. Where there are signals here, there are markings too, I have yet to see that not hold. The progression is not only subjective, but firmly rooted on the ground. OTOH, people here debate, whether unmarked crossings can be mapped with certainty, especially mid-block ones, but not only those. Perhaps the orthogonality lies in what can be taken for granted and what should be special-cased?

Usage of marked and uncontrolled overlaps here one on one, as far as I observe. The value chosen is seemingly due to editor templates.

PS: The exact wording of the code of law that creates crossings where there is a crossroads might be helpful too? Or does this stray off-topic too much?

PPS: What I wrote about iD users is just an account of what I got told; In my personal observations, iD users are just as diverse as the openstreetmap community as a whole.

[tarunik]

As a fellow North American mapper -- I do not use crossing=uncontrolled. This is partly because a marked midblock crosswalk (or a passive rail level crossing, for that matter) may have controls that are not equivalent to a crossing=traffic_signal or other active control means (such as a rectangular rapid flashing beacon system on a crossing=traffic_signal, or simple highway=stop controls on a railway=level_crossing). Furthermore, the US Manual on Uniform Traffic Control Devices is worded in a way that permits any state or local authority in the US to omit crosswalk markings at any intersection approach that is controlled in some fashion, consistent with @1ec5's examples. (The crosswalk markings are only required to legally establish the crosswalk at midblock locations -- intersection crosswalks are defined automatically as a function of state traffic law, and may exist even in the absence of traffic controls at the intersection.)

[1ec5]

By the way, it isn’t terribly difficult to find unmarked, signalized crossings in Mexico (a Vienna Convention country) and South Africa (which is influenced by the Vienna Convention), though they probably are more common in the U.S. and other MUTCD-influenced countries.

[1ec5]

PS: The exact wording of the code of law that creates crossings where there is a crossroads might be helpful too? Or does this stray off-topic too much?

We are a bit off-topic for this presets project, but it is an engaging discussion. 😄 Here are some examples of U.S. state laws establishing crossings at intersections:

• California – Section 275(a) is very broad, including many unsafe crossings that most laypeople (and probably the police) would regard as jaywalking. For example, it technically allows crossing a busy four-lane road at a T-intersection with a smaller street, where there’s no crosswalk markings and only the smaller street has a stop sign. It also allows crossing on the wrong side of an intersection, opposite the side that has a marked crosswalk, unless otherwise signposted.[^noped]
• Missouri – Section 300.010(8) says a crossing is implied by the sidewalks on either side of an intersection.
• Ohio – Section 4511.13(3)(a) specifically provides for unmarked crosswalks controlled by traffic lights intended for cars.
• Texas – Section 552.001(a) does the same.
• Most states have adopted some variation of this portion of the Uniform Vehicle Code. Section (A)(3) is about pedestrians using the traffic lights for cars.

These statutes are primarily about legal rights of way, which becomes relevant in case of litigation. The more relevant authoritative documents are the engineering standards published by state and local departments of transportation, or the influential NACTO recommendations, but mappers are very unlikely to consult these documents anyhow.

[^noped]: In urban areas, most such crossings are prohibited by or signs.

[hungerburg]

I skimmed the linked U.S. state law and code. The OSM distinction between marked and unmarked crossings directly mirrors the language I found there on crosswalks. In my country, the law only knows marked crossings. Crossing the carriageway at an intersection is in no way different from doing it anywhere else. There are either no unmarked crossings, or there are a gazillion. (I personally mapped midblock unmarked crossings, where there are lowered kerbs or other indications.)

I do not know, where the "uncontrolled/signals" proposal originated, but if it was in a place like mine, I'd fully understand, why the question "But, does it have signals?" was considered so important, as to turn that into a value for the crossing key, and "unmarked" only came as an after-thought; while in the U.S. perhaps the question, "But, does it have markings?" might be considered the decisive feature.

Hanging on this thought, on orthogonality, and maybe why this issue is still open, last but not least, because this is a tagging-schema repository:

1) A matrix is more capable than a single vector, in order to formalize complicated subject matter. From an engineering point of view, that sounds quite reasonable. The composing vectors in the matrix do not need to be orthogonal. 2) What then is a bare highway=crossing? Because for some, where there are neither markings nor signs, there is no crossing either; while for others, crossings abound, c.f. #8463 - Will we have to introduce highway=intersection too?

PS: The picture from Valladolid clearly shows a place in slow transition: There is an older crossing up the street, where the kerbs are lowered and the carriageway has a pinkish tone. The pedestrian signals that mark the crossing in front, themselves do not conform to the convention too. Also, on account of a visitor, they are not operational and people look at traffic and vehicle signals instead.

Update: Scene with disfunctional pedestrian signals, https://youtu.be/p-p6B8-EZ_A?t=167

[1ec5]

What then is a bare highway=crossing? Because for some, where there are neither markings nor signs, there is no crossing either; while for others, crossings abound, c.f. https://github.com/openstreetmap/iD/issues/8463 - Will we have to introduce highway=intersection too?

I think we’re actually on the same page there: https://github.com/openstreetmap/iD/issues/8463#issuecomment-1066069238.

[matkoniecz]

crossing=uncontrolled and crossing=marked represent two orthogonal tagging schemes

+1

If anyone cares about actually fixing the root of problem we need to find way to tag this things separately. Trying to fit it in one tag is not going to work well. Otherwise we will go in circles and produce data that is not actually so useful.

[matkoniecz]

And as far as my recommendation for iD go: stop supporting any value of crossing and stop offering to tag it in the presets and wait until the mess is fixed.

[matkoniecz]

I do not understand the bit about unmarked yet signalised crossings. I myself map pedestrian infrastructure, not paint on a surface.

The problem is that for various reasons (for example legal implication, mapping from aerial imagery, 2D rendering etc) some people want to map road markings ending with various kinds of issues.

For example using marked/unmarked based on aerial imagery without checking whether signals are present there.

[AntMadeira]

As an iD translator in Transifex I recently noticed that highway=crossing + crossing=uncontrolled was added to presets. I don't seem to see from this thread any decision to do so, but worse than that is the description "Marked Crosswalk", which is exactly the same as highway=crossing + crossing=marked. I think this is a terrible way of contributing to the resolution of this old problem in OSM tagging. Shouldn't that preset be removed until we reach a better solution instead of throwing more gasoline into the fire?

[tyrasd]

That string was added in 3525fe694d931ec90c66b97b2fd2a84c05672854. Note that the added preset was hidden. The main goal was to render icons also for the crossings which were mapped using the "traditional" tags (which might have been mapped by a JOSM user), see #390. A side benefit was that now these objects also got a matching preset, showing the dedicated preset fields in the user interface, rather than falling back to the "generic crossing" preset as it was the case until then. Note that this preset was created as a hidden one, meaning that only already existing map features were affected by it and iD users were continued to only be offered the preset which used the crossing=marked tag. I hope this clarifies that this didn't worsen the situation. But if it did, I'd be glad to learn how this could have approached differently.

Please also note that there exists a proposed changed to iD's tagging presets in which would change iD to not use the crossing=marked tag anymore for the "Marked Crosswalk" preset, but to instead use the crossing=uncontrolled tag. Maybe you would like to take a look at the pull request #590 and leave your opinion about it there.

[AntMadeira]

Thank you for your reply and explanation, @tyrasd. I didn't know that the added preset was hidden. That makes sense with the further explanation you gave for this decision. For editors, this is no big deal, because it's not searchable, but for translators it should be possible to add side notes to important changes in Transifex presets so that we have more information on how to translate them correctly. Mind you that I come here to participate and learn about iD, but I would argue that most of iD's translators don't even know where translations come from in the first place. It's a complex world to navigate...

About the proposed change, I would rather see a structure with crossing=marked/unmarked with the keys crossing:signals and crossing:markings to complete the data. In my opinion, uncontrolled is too ambiguous and error prone. But this is a ship that may already be far gone...

[1ec5]

For editors, this is no big deal, because it's not searchable, but for translators it should be possible to add side notes to important changes in Transifex presets so that we have more information on how to translate them correctly. Mind you that I come here to participate and learn about iD, but I would argue that most of iD's translators don't even know where translations come from in the first place. It's a complex world to navigate...

This is similar to ideditor/schema-builder#27, which proposes a mechanism to allow id-tagging-schema to mark a string as not for translation. That could be a special case of providing more prosaic background context about a string, particularly when the American English term is easily misunderstood by translators.

It is confusing that a single Transifex project contains translations for four different repositories (also editor-layer-index and osm-community-index). If we ever move away from Transifex for openstreetmap/iD#7508, migrating each repository piecemeal would lessen the disruption.

About the proposed change, I would rather see a structure with crossing=marked/unmarked with the keys crossing:signals and crossing:markings to complete the data. In my opinion, uncontrolled is too ambiguous and error prone. But this is a ship that may already be far gone...

Don’t give up hope yet! This is similar to the idea in #507. @tyrasd has expressed openness to deconstructing the crossing presets in favor of these fields but just needs some extra convincing through the proposal process. 😉

[AntMadeira]

Recently, more strings were added to Transifex which contribute even further to the confusion regarding crossings. To illustrate what I mean (and these are only for untranslated ones):

[hungerburg]

In my area _deAT these might translate as:

• 1,3,4,6 - Schutzweg
• 2 - Radfahrerüberfahrt
• 5 - not translatable, has no legal base, but may be mapped of course, if e.g. lowered kerbs or traffic calming islands suggest this as a preferred place to cross

[1ec5]

Related: #590 #658.

[Dimitar5555]

It seems like this issue can be marked as resolved. Currently "Marked crossing" preset uses crossing=uncontrolled + crossing:markings=yes.

[riQQ]

This was changed with #590 / 80c5df168a20e7df872d23d5f5ea2629b9489dec

Direct anchor link to relevant file: https://github.com/openstreetmap/id-tagging-schema/commit/80c5df168a20e7df872d23d5f5ea2629b9489dec#diff-41416f8f17117f5b868d3e892541a6594867313c6fef7841d39a2027bed19bc8