Thatās not the point, mistakes happen all over OSM. The thing is that in some countries itās unacceptable to have a highway category that doesnāt reflect the national classification while with importance everybody should be able to adjust it based on what function the road fulfills in the road network.
Itās not that difficult to look at importance and service tags.
I have no idea how you got to this conclusion. Do you mean importance describes only the importance for cars? In that case youāre right that I forgot to mention that there should also exist importance:hgv, importance:bicycle, etc. but only the regular importance would be rendered on default layers while importance:bicycle could be utilised by cycle-focused maps.
Yup, the current scheme only supports tagging importance on roads but not paths. With the new importance tag you can add it to cycleways (i.e. replacing network=lcn/rcn/etc) and also footways. The importance for bicycles on roads would be tagged as importance:bicycle.
Well, replacing the network tag with a new importance tag is an absolute no-go, for the same reasons that deprecating the highway-values with it is.
Furthermore, lcn/ncn/rcn networks usually require the presence of a route marker and an operator both of which fulfill the verifiability requirement. Verifiability of āimportanceā sounds to me either completely subjective, or directly derivable (especially in the case of paths and cycleways) from physical features (designated, segregated, surface, smoothness, etc.) and/or its place the routing network (i.e., if a thereās only a single/shortest contiguous cycleway between any āimportantā points A and B, routers will find that fact easily without any further tags).
Cycleways donāt usually form contiguous routes between many places (ncn routes often go through roads as well as cycleways), and whenever they do, those are easily identified directly in routing graphs.
Yeah, youāre right that some of the current software uses this tag so for the time being it would also have to be double-tagged. But I have seen many people point out how nonsensical the usage of network=* in this context is.
With cycleways itās probably best to just copy the official classification. Importance diverges from official classification a lot more with car roads.
You seem to feel that some of the problems with current tagging come from its origin in British road classifications, but it feels like this risks replacing it with tagging inspired by Poland, Germany, and a few other countries, which can hardly be expected to produce a better outcome.
Even in Europe, itās not clear how this would work for Ireland, for example. Currently roads between main cities are tagged as trunk, regardless of their physical construction. The next level of importance is tagged as primary. (These reflect official classifications in practice, but the official classes broadly reflect importance in the overall road network in any case). It appears that your proposal omits one whole level in the hierarchy, and would force these into one importance level? Or else force a downgrade of lower levels, which would cause problems at the lower end?
We are trying to help you to structure your proposals better. You ask people to read a lot of information, then say ābut Iām not proposing that, letās just talk about thisā, forcing the reader to reread the 1st part to try to work out what is relevant to the 2nd part. I would suggest only writing about concrete proposals that you are certain you want to discuss.
I donāt like the idea of moving all these highway types to one top level tag to start with, but if it had to be done, maybe function=? I think that doesnāt come with much baggage from other tagging schemes but I havenāt checked taginfo.
Maybe there is a misunderstanding of āmistakeā here, but, surely the whole purpose of the new tag is to have a different outcome to the current scheme?
If most people in a mapping community feel that importance is best mapped by reference to official road classes, they will still feel that about the importance tag. Unless, perhaps, there is very clear guidance about how to objectively define importance, which I donāt see anywhere in the proposal or links.
A good example of my point about structure. At some points you seem to accept this wonāt happen and itās not part of the proposal, but here you answer a doubt by saying it is proposed?
Most highways discussed here have only one surface value (=asphalt), and most are at least two-lane. Furthermore some highway classifications are based on signage (so analoguously covered in the ncn, etc network tags). So Iād say the analogy is far from perfect.
Also, aside from this thread, thereās another thread with (currently) 107 posts about a user changing the highway-value of a road. Would you think it a good idea if we were to have similar edit wars with the importance value of cycleways? For physically exceptionally good cycleways, we already have the expressway-poposal.
Another thing that you didnāt address was the fundamental difference I mentioned between roads and cycleways. Cycleways donāt, usually, form a dense network worthy of complex classification simply because there are very few (or, indeed, no) contiguous cycleways between far away places.
Itās not a replacement but rather a merging. In the UK there is some usage of expressways=yes but but itās used on regular tags like highway=primary or highway=trunk while in Germany there is highway=trunk used just like I want expressway=yes to be used but thereās no highway=* value for those most important non-motorway roads and just =primary is used.
I would like for the trunk road network in Ireland to stay mostly the same, though the UK probably overuses it. Either way, I think trunk roads should be more-or-less evenly distributed across continents (or islands, like Ireland or the UK) so that the network looks rather coherent and not with the trunk equivalent importance cutting off at Cork unless the consensus is that cities like Tralee are both too small and uninteresting for tourists.
Well, what Iām proposing to start using today requires just taking away retagging to highway=road which is simply mentioned as the end-goal to fix the current mess with highway tags. Everything else applies.
That will require, or at least it would be nice, to make it consistent with railways also but railways already have a tag that uses the word function. But is that really worth it? Function and service are quite similar words and service has historically always been used in OSM for numerous features i.e. highways, railways, even waterways and more so I think we should stay with this tag. Weāre supposed not to change things too much after all so if we can keep driveways with the same tagging, then we should do it.
Itās to have a consistent scheme across continents and the entire world. In Lithuania, a paved road is enough for secondary and unpaved roads are tertiary or lower which is just not a good way to make highway classification.
I mean, thatās fine. At least it will still prevent switching highway values at national border crossing.
In my opinion it should be a city population threshold which makes roads be roads connecting two cities above this threshold be able to be tagged with a certain category. This threshold should be chosen based on what will make the majority of the ways of the parallel national classification correspond to this category with singular exceptions and there could also be exceptions when e.g. a city is not very large but lots of tourists visit it. To give an example, I proposed to use highway=trunk in Poland to connect cities above 270k population (coincidentally thatās the 10 largest cities also) which makes it line up very well with the TEN-T network. Communities should also base their classification with the importance value on the other side of the border. I think Europe has it easy with the TEN-T- and E-networks while other continents should figure out their E-network equivalences or try connecting the biggest cities and agglomerations.
To repeat my original post, for now we tag importance, service and other tags like expressway, motorway, living_street, highway:category and in tens of years, when most software supports this scheme, all/most roads get changed to highway=road, similarly to how it is with railway=rail.
Yup, importance is somewhat subjective just like the current highways.
Wait until you hear about what goes on in the Polish communityā¦
To actually solve a problem which causes a long discussion, people should compare different features of the road such as itās parameters, traffic, role in the network and presence in networks like TEN-T, official classification and local knowledge. And if everything fails, democracy should decide.
Welp, thatās great news, no? Itās a simple move from network to importance so that network can be used to tag other things, similarly to brand and operator keys.
The N21 and N22 to Tralee are both tagged as trunk already. Some other roads around Tralee are tagged as primary, such as the N86 heading west. How would your scheme preserve this distinction? Are you saying the N21 and N22 would become āinternationalā?
Again, how does moving to the importance key change that? Can you give a specific example of a road that would be classified differently, and explain why that can be done with the importance key but not the existing highway key?
It seems that Tralee is an important destination, probably because of tourism. So once a destination is considered to be of this level of importance, it should be connected to other places of this importance by a road with this importance=* value. The other roads arenāt the best roads to get to Tralee from big cities like Limerick or Cork.
Half of Europe doesnāt have the extra importance layer of trunk, some countries copy the national classification or pavement of roads and sometimes donāt allow for values based on actual importance, some countries have unclassified be the same as residential but with different surroundings.
The highway key is ambiguous which has made many countries have their own classification, oftentimes disconnected with their neighbours. Someone mapping in one country needs to read on the wiki how to classify roads in the neighbouring country if he wants to map there.
Probably by just copying the criteria for network=icn/ncn/rcn/lcn.
Coming back to this, I now realise I might have made a mistake with the importance values and they might be off. Arenāt trunk roads of national importance and primary roads of regional importance actually? I was basing it on the Polish classification in which national roads are usually highway=primary and regional roads are usually highway=secondary but seeing Wikipedia, the Polish translation for the page Trunk road is National road. In this case I think importance=international would be used sparingly, only for roads connecting capitals (thatās how it is in the original importance key proposal). I donāt know what should be the usage of this tag in the US though.
icn is a EuroVelo or other international signposted route. ncn is a route signposted as part of a national cycle network. rcn is a route signposted as part of a regional cycle network. lcn is a route signposted as part of a local cycle network.
I guess you should just do the same as you do with cars so you check if the roads and paths are part of an important connection but here you especially need a lot of local knowledge and surveying. But just by looking at the map, this particular road doesnāt seem very important.
Well, as always, the problem is that itās not clear where the dividing line between national and regional importance lies - especially in countries that donāt really have administrative units that could be classed as āregionsā.
In any case, for Ireland, the official classification recognises two levels of ānationalā road and also āregional roadsā. All types of national road have N numbers but are clearly distinguished from each other officially, e.g. they have different default speed limits. Regional roads have R numbers. These 3 classes are mapped as trunk, primary, and secondary in OSM. The legislation simply lists the routes rather than explaining the classification, but Iād say it is roughly: (1) roads connecting cities to each other and to large towns (2) roads connecting large towns to each other and to smaller towns, and important tourist routes (3) roads connecting large towns to the surrounding area.
Thatās why I feel your proposal misses a level. Probably only the road from Dublin to Belfast would be promoted to āinternationalā, so there are still two levels of importance higher than āregionalā.
(To answer your earlier question, Tralee itself is not a major tourist centre but nearby Killarney is. The N70 is a good example of a road that owes its ānationalā classification, and primary status in OSM, to tourism. You canāt use it to get to any significant population centre, but it is part of the āRing of Kerryā which is a very popular driving and tour bus route.)
You keep repeating the inconsistencies between countries with the current system. But I donāt think anyone is disputing that. The question is, how will we avoid ending up with the same inconsistencies with the importance tag? And I donāt see that you have answered that anywhere.
In summary, leaving aside specifics like link roads: if we end up with a 1-to-1 relationship between the current main highway= tags and the importance= tags, all we will have achieved is decades of dual tagging for no gain. So this only makes sense if the mapping moves away from 1-to-1. Can you give specific examples of how and where you think this will happen, and how the relevant countries will accept maps that look different to what they are used to?
It would just be about connecting bigger cities, I guess.
Another word is āsupernationalā which I guess is the best word to describe the intended use of highway=trunk and solves the issue of usage in the US.
I think it should be pretty clear that importance is strongly connected to admin_level so which admin_level values do you think should have corresponding importance values? Or do you think there should be something like 1.5 and 2?
What inconsistencies? What it solves is making clear guidelines for importance after highway lost its original meaning. Currently, all countries have consistent tagging schemes when looked at individuallyāitās just that the usage has diverged a lot from the initial designation.
When looking at individual countries, yes, but looking at it globally, itās a huge step forward with coherent road network tagging. Thereās also local benefits in e.g. Lithuania by gaining importance tagging in all areas and even half of Europe gaining the trunk-equivalent importance. Other benefits include differentiating highway=unclassified and =residential.
I wanted to make as little changes to that as possible because renderers and routers depend on the current scheme and so moving to the new one shouldnāt require making up new values for recommending roads of new/changed importance or thinking about using different colours for them.
I believe that this half of Europe will start using the trunk-equivalent importance with the current trunk roads being replaced with expressway=yes, little issues with unclassified & residential or living_street will be resolved, weāll have a highway tagging scheme that is consistent with the railway & waterway schemes and last but not least the importance key will finally get some love and be used for highways but also railways, waterways and other stuff too.
So really this will have the biggest impact on Europe and countries that utilise trunk for access control. But as a recompensation, these roads, tagged with expressway=yes, will have a special rendering, similar to the one on americanamap.org.
That is NOT clear to me. It would never have occurred to me, in fact. Can you explain how road importance is connected to admin level?
(To me admin level is only really relevant within its own context. It reflects arbitrary structures that can and do change over time, sometimes drastically. It has very little to do with real world concepts other than administration, like the places people live and how they travel between those places).