Talk:Highway link

Naming links

How should links be named (to avoid them showing in red in potlatch)?

Don't tag for the editor. If the link has a name, use that. If it doesn't, don't enter a name. (Roughly) all alerts of the data not complying with that/each editor's rules are notices only. Alv 22:23, 22 October 2010 (BST)

If buggy validator demands names on links - it is a validator bug. Mateusz Konieczny (talk) 10:24, 15 February 2021 (UTC)

Resolved: Potlatch 2 is dead, if new version after decade has the same bug - report it to whoever maintains it Mateusz Konieczny (talk) 10:27, 15 February 2021 (UTC)

Highest value

so the admonition against tagging for the renderer does not apply here.

Everybody can stop reading after that half sentence. This is a rendering problem, so fix the rendering rules, not the tagging. I thought that this was clearly stated in the original mail discussion. --Cartinus 17:16, 2 July 2010 (UTC)

What's the rendering problem? What's being said there is that whether it's tagged (for instance) primary_link or secondary_link makes no difference to anything except the renderer. --NE2 18:48, 2 July 2010 (UTC)

To render a link properly, it needs to start at one 3-arm node and end at another (3 or more). You get rendering artefacts if it ends at a 2-arm node (unless the two arms happen to be the same colour). The other thing you need to do is to put the link under the two 3-arm nodes at each end. So fine, you might think, just render links under everything (this is what Mapnik currently does). Unfortunately you need to render them on top of any intervening 3-arm node (eg a service road coming out in the middle of a link). The least amount of information you therefore need (to render it neatly) is the lower of the two classifications that the link joins. You render the link just under the lower of the two classifications, and any intervening even-lower classification road goes under the link. If you would prefer the link to be the colour of the higher of the two classifications that the link joins, you also need to know the higher-of-the-two classifications. The higher-of-the-two-classifications is (per the wiki) used as the basis of the highway tag (though this doesn't always happen, because it looks a mess unless it's reasonably symmetrical). Maybe the lower-of-the-two-classifications could be recorded with a links_lower tag. It would be simpler if we recorded the lower classification in the highway tag, and the higher one in a links_higher tag, but maybe life isn't going to be that simple. If you add both links_lower and links_higher tags, then a bot can sort it out one day.--RichardMann 22:30, 2 July 2010 (UTC)

If I'm understanding this correctly, the issue you have is the renderer drawing links on top of lower classification roads. This could simply be fixed by the renderer drawing all links under all roads that aren't links. Fearsyth (talk) 15:21, 2 November 2019 (UTC)

Hi All,

I would like to discuss the current implementation of using the highest highway=*_link route possible for link roads.

Using this method there is inconsistent application. Particularly at turning segements which are also marked as link roads but commonly have the lower classification marked. It makes practical sense that it be the lower classification. But that's a big editing task (May be assisted by bulk editing which is easy enough) What's your opinion guys?

Using this method a car magically becomes a part of the highest route even though only one car may travel on that road per day, if it goes into a motorway, its a motorway link. The lower route makes the most practical sense and is what is used by Google, HERE and the state and national government mapping (Which records the official road classifications and heirarchy) of my area (Australia & Western Australia).

Additionally the current justification for higher classification - that link roads have the same attributes or restrictions as the higher classification road is untrue. For Australia wide (I have confirmed this information) it is a typical standard (I can't say for every single situation) that the speed limit remains that of the lower road classification until the slip road terminates. Though speed limit signs may be 150m or so prior to the actual "merge point" there exists no road here in government mapping, that is to say that 150m or so prior to the merge point the slip road is the main road. It is also at this point where you are on the main road that it becomes illegal to have horse drawn vehicles or mopeds or bicycles for example, depending on road. It is not illegal to have these on the link road (again can't confirm every single situation). Another note is that motorway links (controlled access highway slip ramps) can still have driveways or roads leaving and entering it provided there is unimpeded traffic flow whilst a motorway cannot.

As I'm exhausted at the moment XD I'll refrain from going into further detail.

Some clarification examples include:

[1] - [2]

12:00, 3 Feb 2015 (AWST)

Add comment here, include time and date. --Aaronsta 10:56, 3 February 2015 (UTC)

Hello ! I agree with you, I would tag an interchange between a motorway and a primary road as highway=primary, link=yes, main__link=motorway, legal=motorway, or an interchange between a non-motorroad trunk road and an unclassified road as highway=unclassified, link=yes, main__link=trunk, legal=classic. Actually a link road should even have its own classification, I got one example where one link road of the interchange is rarely used (it allows to go backwards because the next bridge is low). --djakk 10:00, 17 September 2018 (UTC)

I did a lot of map comparisons recently and it is true that Google, Here and Waze prefer to classify links by the lowest classification of the two linked roads rather than the highest, although they are all often inconsistent. In particular, Google prefers to classify by the highest on major highways, but only when the access ramps are very distinct, not small nor of an informal appearance coming from city streets (this is particularly common in Brazil, where the road system is not so well designed). The article mentions that the Russian community has gone the other way, classifying the links by the lowest class, and it usually appears more organized and readable on the map, like on the Moscow Ring Road - Borovskoye Highway, Ozyornaya Street interchange (comparison). Semantically, I think that direct links between major highways (usually high speed, except for U-turns) are different from links between major highways and city streets (usually as slow as the city's system, if not even slower). Based on the default speed limits, a lower classification would normally make more sense in the second case (link between major highway and city streets). For example, the speed limit is, by default, unlimited in motorway links in the United Kingdom, which is somewhat reasonable on the major links of a large interchange like the M1 - M25 interchange, but very far off on the access ramps of a minor interchange like the M1 - A405 interchange (comparison). Based on the default access restrictions, it may make sense to adopt the highest class only in links to/from highway=motorway. I think it helps to reason like this: if "link" was not a special type of way in OSM and we had to use a normal type, what classification would links have? Based on the examples given by Martin Koppenhoefer (Dieterdreist) on the tagging mailing list, I think they would normally be the lower class of the two linked roads. --Fernando Trebien (talk) 13:42, 20 February 2021 (UTC)

Y Junctions should NOT be classified as links

Observe this Y junction: [3]

2-way roads that split upon arrival of a crossroad should not be classified as links.

My reasoning is as follows

1. For a highway to be a link, it must branch off another highway that is of the same direction of travel. If a Tertiary Road ends in a Y Junction abutting a Primary Road, you do not have an option to take another route at the split. There must be options in order to classify something as link.
2. Routing is negatively affected by classifying them as links. A GPS giving turn by turn directions will state "Take the on ramp ahead and follow it onto the Primary Road" Since they are not thought as to be on ramps this can confuse travelers.
3. Each branch of the Y Junction is still considered the road which it split from. The name of the road should be added to the branches of the Y Junction. Therefore, it is not a _link because _link's do not have names

It seems there is a majority consensus on this based on what I see in OpenStreetMap. However, the picture on the _link page shows Y Junctions being classified as links (???)

If those Y Junctions are links then that would mean, for example in the picture on the official page, the Secondary Road (that turns into a Y Junction) would become a link at the previous node where two Residential Roads intersect the Secondary Road (which makes an elbow turn). BECAUSE at that point, there is no other option other than to merge onto the Trunk Road.

Let me hear your rallying cry and we can settle this! — Ranger444 ^(talk) 21:38, 21 September 2017 (UTC)

Wrong. Links are genrally short highways that connect one road with another one, most of them are in fact unidirectional, they have no return from what they come (U-turn is disallowed once you start taking the link).

Also there's nothing in a link that says it is a "ramp". Even if this is the only direction you can take your router will just say "Taking the link and continue on rpad to..." but may still allow before taking it to instruct you to make a u-turn before taking the link.

at end of the link you may still have an option: the target road may be bidirectional or not and you may have to merge or not depending on the number of lanes on the target road, but you won't merge on the link itself, except if the link itself as a reduction of its own lanes (e.g. 2 lanes at start, only one lane before the junction at end of the link).

You're just alone with false reasoning. You are the only one confused.

Your example fits perfectly what are links: the 2 small branches of the Y connecting the north-south street to the west-east road are unidirectional links. The number of unidirectional lanes in the same direction on them does not matter, but what is important is that you must not stop on it except at end, and cannot make a u-turn on it. — Verdy_p ^(talk) 05:24, 8 December 2017 (UTC)

1. "connecting the through carriageways/through lanes of highways to other roadways of all types" If there are not through lanes which you can branch off from, then it would not be considered a link. This is from the first sentence on the link page. Please thoroughly read the wiki before correcting others. 2. Getting onto a link must always be an 'option' not an obligation. Meaning, if y-junctions are to be classified as links, then you should begin the link classification from the last intersection (wherein exists the last option to turn off and not take the link). Another reason for this is that highways don't randomly turn into links without encountering some sort of junction where you would have an option to go in 2 or more directions. Take this Y-junction as an example [4] Per your faulty reasoning, the residential road would be classified as a link starting at the service road near the clinic. Which brings me to my final point 3. Classifying y-junctions as links yields to an incredibly ugly map. Aesthetics IS important. Y-junctions are minor connectors that do not offer different directions of travel. They are merely a safety feature to avoid driver's from taking up the whole road and thus blocking traffic from incoming cars. It appears you are the only one alone and confused. — Ranger444 ^(talk) 15:21, 8 December 2017 (UTC)

U-turns permitted before the tow oneway Y branches means a different topology: in yiour case you have a normal street, finally terminated by two oneway links with no U-turn and no stop (you cannot block the other vehicles. So you're still wrong here. The two branches in your example are really links... — Verdy_p ^(talk) 15:37, 8 December 2017 (UTC)

Ok then with the no U-turn you have no prior option to turn off but the service road near the clinic. Thus, per your reasoning, the road would be classified as a link from the service road because a road cannot randomly become a link without a junction that has turning options. This is obnoxious and ugly. Address this issue. — Ranger444 ^(talk) 15:51, 8 December 2017 (UTC)

In fact the link includes the full Y up to the intersection to the north, so all 3 branches are links! — Verdy_p ^(talk) 16:10, 8 December 2017 (UTC)

Ahh, thank you for illustrating the ludicracy of your logic. I believe all can agree with me that having such minor roads be classified as links for that great of an extent is both ridiculous and ugly. — Ranger444 ^(talk) 16:30, 8 December 2017 (UTC)

Once again you are using flaming words: this voids all your arguments. You are just starting to contest isolately something that has largely been approved collectively and used without any problem by our millions contributors. Good luck but OSM does not work this way. You cannot contest things alone (and no this is not jsust "my" logic: you are insulting every other people as if they were considering this only as a joke or game, when what they do is serious and they have made many compromizes to adapt to a complex landscape of situations. Apparently you don't seen to understand that cartography has NO rule, only some known best practices, always with exceptions EVERYWHERE you go. — Verdy_p ^(talk) 18:13, 8 December 2017 (UTC)

Ok, I'm sorry for my 'flaming words'; however, you were the first to be disrespectful. I understand that cartography works this way and that rules and guidelines are under an ever-changing landscape. This issue has not been talked about on this page before and I saw it as being pertinent. The thing about an ever-changing landscape is... that it changes! I saw what I perceive as a broken chain of logic. I want to fix it. During our discussion it is shown that my new rule has plenty of merit. I would like to hear some other users opinions on this however I think we will have to wait a few months for anyone to comment. Once again, it is unfortunate that our discussion took a rude tone, I apologize for my part in it. — Ranger444 ^(talk) 14:18, 11 December 2017 (UTC)

No you were the first one to be disrespectful by using the terms "ludicracy of your logic", and even before with "your rallying cry" against the "majority consensus" that you wanted to denounce. You've been flaming since the first post in this thread. — Verdy_p ^(talk) 20:52, 14 December 2017 (UTC)

The whole point of the talk tab is to parse out things that the majority might do that users see as flaws. Also, 'rallying cry' is not a disrespectful term, I was just trying to add some energy to an otherwise boring topic. furthermore, your logic was ludicrous in my opinion, so I guess I'm sorry if I offended you. Now, as I have previously apologized, can we get back to the issue? Do you not agree that in the stated example changing the dual carriage-way AND the two single carriage ways into links is both ridiculous and ugly? In OSM we need rules that make sense. I stick to my proposal. — Ranger444 ^(talk) 21:13, 14 December 2017 (UTC)

I strongly agree with this proposal. All roads have a way to "enter" and "exit" to them. Therefor a Y-Junction should not be classified as links, unless it is a ramp into a Trunk/Motorway. Please consider the thought of an entrance and exit to a road, what is the difference of it being split into a Y-Junction or a normal 2-way connection, if you think Y-Junctions should be links then you should also be making that last section of a way a link regardless of it being a Y or not. This also affects navigation, links do not have names, therefor you would have a "turn right" instead of "turn right into Maple St" for an example. This is common sense, and none of the posts have been disrespectful towards the community. I stand with the Y-Junction NOT as links. — Thenodifier ^(talk) 21:31, 14 December 2017 (UTC)

Sometimes, in written media, it is difficult to distinguish between toxic sarcasm and friendly play, so let's try to assume good faith and avoid grandstanding. Since I'm late for the discussion, I'm just here to point out to future readers the examples given by Martin Koppenhoefer (Dieterdreist) on the tagging mailing list, from which I also argue that Y-junctions should not be classified as links, they belong to the minor way from which they branch (also getting its name) and are just a physically divided section of that way. --Fernando Trebien (talk) 14:48, 20 February 2021 (UTC)

This section become clarified to match actual mapping practice (that makes sense). Mateusz Konieczny (talk) 16:55, 21 September 2021 (UTC)

Turning lane changes

Hello, the "Turning lanes" section is a little confusing and seems to disagree with the rest of the link policy, and doesn't match with most mapping I've seen. When turning lanes connect to a higher classified highway, the second sentence contradicts all of the classification policy defined on this same page. It looks like it was last edited in 2013 without any discussion I can find. I propose editing it a bit to reflect the rest of the policy and make it more clear. Specifically, replace the second sentence with "They should be tagged with the _link classification of the highest classified road they connect to”, replace the example shown in Note 5 with the example found here, and update the topic header to "Internal Turning Lanes" to better clarify what feature this covers. Thanks. --Awiseman (talk)

New image

Hi, I think the current image isn't the best example of a highway link, so I am thinking of changing it to this. Thoughts? -- Awiseman (talk)

Resolved: done Mateusz Konieczny (talk)

Roundabouts

Should roundabouts (the circular way in the roundabout, in particular) be tagged as links? I've seen this variously on OSM. It seems to me the general practice in printed maps is to tag them as non-links of same type as the most important road connecting to the roundabout. Rostaman (talk) 04:01, 18 January 2020 (UTC)

Central roundabout way would not be downgraded to link, though it can be in theory link road on roundabout formed by junction of link roads Mateusz Konieczny (talk) 16:56, 21 September 2021 (UTC)

Different tables of connectivity of highway links

Hi

On this site https://wiki.openstreetmap.org/wiki/Link_roads_between_different_highways_types we have other connectivity links between types of roads than on this site https://wiki.openstreetmap.org/wiki/Highway_link ( table on picture)

There is some inconsistency on tables e.g. trunk to service link connection should be service and service to trunk connection should be trunk_link.

Best Regards Ewelina - Evibes (talk) 11:19, 17 July 2020 (UTC)

I've removed the image and pointed to the article to remove the duplication. The article seems to be more up to date, though perhaps not very correct according to the discussions it points to. The image did not include the table caption, which suggests it is a personal point of view, not something widely adopted. --Fernando Trebien (talk) 21:45, 19 February 2021 (UTC)

Definition of Channelised

Updated the wording so it can include physical and other separating features such as barriers or paint. Many slip lanes with their own paint will often have their own traffic control device. The presence of its own traffic control device indicates its physical separation.

On top of all the controversies, should highway=*_link really include non-phsyically separated ones when highway=* are only splitted when there's physical separation? You could do the same for non-physically separated ones with turn:lanes=* easily (without the need to fiddle with many geometry), and Relation:connectivity if complicated. There's an edit series going on from at least https://wiki.openstreetmap.org/w/index.php?title=Highway_link&type=revision&diff=1869950&oldid=1818573 to https://wiki.openstreetmap.org/w/index.php?title=Highway_link&type=revision&diff=2067085&oldid=2005581. ---- Kovposch (talk) 09:17, 5 February 2021 (UTC)

Crosshatch markings on Boulevard de la Pétrusse, Gare, Luxembourg

turn:lanes=* is almost always the best representation when there is no physical separation, for example, when the entrance or exiting lanes are separated from the main lanes by a line (continuous or dashed). However, channelisation marks such as chevrons and crosshatchs could be treated as analogous to physical barriers depending on local law. Driving on them on entrance and exit ramps is probably forbidden in most countries. Where I live (Brazil), chevrons and crosshatchs sometimes replace a median strip in some arterial streets, and in these cases it is forbidden to drive over them and turn left to reach adjacent properties, so in practice they act as a physical barrier. Apparently this is not the case in New Zealand. In the United States, they may be used in short stretches of turning lanes in which case I wouldn't consider them analogous to a physical divider either. In the main example of road surface marking on Wikipedia, it probably does not act as a real divider (too short, not that wide), but for a longer stretch it would depend on local law (in this case, the law of Luxembourg). --Fernando Trebien (talk) 12:11, 20 February 2021 (UTC)

I wouldn't add link roads for painted islands, because 1) they are usually ignored and driven over by large vehicles, 2) they're so common that they would dominate link roads, 3) they're usually very small and hence hard to draw appropriately if the placement=* for the main roads isn't set to the closest lane, 4) because of their small size they're nearly always in front of traffic lights, so they increase routing complexity without gains to the end user and 5) if the markings are washed out (as they often are), the routing user will wonder where the link road is. Rostaman (talk) 14:35, 9 February 2021 (UTC)

Frontage roads versus highway links

Sometimes the connection between two highways is a way with direct access to private properties, so it works both as highway link and as a frontage road. I think that, in these cases, the way tends to have the speed limit and access restrictions of other city streets, so the way should not be of a *_link type, because for various purposes it is more like a city street than a highway access ramp. What do you think? --Fernando Trebien (talk) 14:35, 20 February 2021 (UTC)

Do you mean direct access to a building/house, or to a carpark? ---- Kovposch (talk) 07:39, 21 February 2021 (UTC)

Usually residences (houses and buildings), sometimes parking lots of office buildings or supermarkets or similar. --Fernando Trebien (talk) 11:11, 21 February 2021 (UTC)