Proposal talk:Narrow width

The proposal says "tag width=narrow is not yet defined at Key:width", and suggests that this should be defined. I disagree - width=narrow should always be discouraged, in favour of a value that is an absolute and verifiable width (i.e. in metres). Alternatively, given that width estimates are always approximate, perhaps it is worth proposing a new way to indicate the precision of the measurement, as in, e.g. width=1m-2m. But if so, this belongs in a new proposal...--Waldo000000 12:39, 18 February 2010 (UTC)

I did some re-writing and gave an example to make this more clearly. --Kslotte 13:02, 18 February 2010 (UTC)

This is either subjective, or redundant.

I don't see this tag as recording useful information. If you know how wide a way is, just add its width using width=*. If you don't know how wide it is, how are you going to say if it's narrow or not? If you add this tag based on "I can see it's narrow", then it's subjective information, and not verifiable. If you set an objective measure for what's narrow (e.g. <2m) then it's redundant information since you could just enter the real width and let the renderer or application decide what to do with the information. Remember: what's narrow for an HGV driver isn't the same as what's narrow for a pedestrian. Jonathan Bennett 13:06, 18 February 2010 (UTC)

The two main point here: 1. define what is narrow, so the render know when to render as narrow. 2. Since "narrow" is subjective we give it approximately definition. --Kslotte 13:22, 18 February 2010 (UTC)

This misses the point I made: If you set an objective measure for what's narrow (e.g. <2m) then it's redundant information since you could just enter the real width and let the renderer or application decide what to do with the information. Jonathan Bennett 15:05, 18 February 2010 (UTC)

It's about helping the rendering; what is narrow from the render point of view. --Kslotte 15:14, 18 February 2010 (UTC)

Don't tag for the renderer. What needs rendering as narrow depends entirely on what your map is for -- large vehicles? Bicycles? Pedestrians? Wheelchairs? No one definition of "narrow" fits all those, so it's better to describe the width specifically, not in vague terms. Jonathan Bennett 15:36, 18 February 2010 (UTC)

Good point. I updated the proposal with your argument. --Kslotte 15:44, 18 February 2010 (UTC)

Note also we have 257 ways tagged width=narrow. We need to find approximately definition for them, so mappers in the future understand what "narrow" really means. --Kslotte 13:31, 18 February 2010 (UTC)

How will you know if the definition you come up with is what those 257 tags mean? Over time, those narrow tags could be replaced with actual widths. Jonathan Bennett 15:05, 18 February 2010 (UTC)

No, we don't know. But if we make some type definition about it; these will get fixed as time go by. As written on the proposal page one option is "Alternatively width=narrow could be made obsolete and force mappers to give width in meters." --Kslotte 15:14, 18 February 2010 (UTC)

Obsolete gets my "vote". Jonathan Bennett 15:36, 18 February 2010 (UTC)

Obsolete gets my "vote", too. Also, don't forget that est_width=* may be a good option if your width value is just a rough estimate --Waldo000000 00:00, 19 February 2010 (UTC)

Already exists: Key:narrow

See Key:narrow. The only thing we need is to improve this page where we say that it is not necessarily temporary. And I would also suggest that "narrow=yes" means between 50% to 75% to what is considered as 'normal width' in local practices (a residential 100% width is not the same distance in Texas or Monte-Carlo). -- Pieren 14:48, 18 February 2010 (UTC)

"get narrower for a brief period of time, such as "under" a long-removed bridge" this map feature has a different approach, even if the rendering should be about the same. --Kslotte 15:00, 18 February 2010 (UTC)

I agree that we already have enough tags to be used for this, from subjective to objective: narrow=yes, lanes=*, and width=*. AM909 17:00, 18 February 2010 (UTC)

Is the norrow=yes something that we want made obsolete or mark it as it should be avoided? Do we have support for this also? Read the definition and tag usage carefully before you give your opinion. Since this is quite similair, we could also take it into the scope and make separate voting for it. --Kslotte 08:53, 19 February 2010 (UTC)

What is narrow?

What is actually narrow? I can for example argue that the road leading to my house is narrow, when cars are parked along one side of the road (which is allowed and normal), than two cars cannot pass. On the other hand, I can also argue that some bridges on the motorway to the next city are narrow, each lane are probably 3.5 to 4 meter wide, but the distance from the edge of the road (white line) to the edge railing of the bridge is less than 15cm, i.e. if I drive with my wheels on the white line I am in danger of scratching the paint work of my car. IMO this type of values should be avoided, and only used if clearer measures are impossible. If a road is truely narrow, take out your measure tape and enter in the observed value. Close to my house is a road that is only 1.5 meter, 2 meter including sidewalks. That is tagged with width=1.5 in the database. This way you can see that you can pass it with your motorcycle, but I cannot under any cercumstance follow in my car. --Skippern 15:09, 18 February 2010 (UTC)

Your point is that width=narrow should be avoided? --Kslotte 15:19, 18 February 2010 (UTC)

Do you have link to the place you refer to? It could be good example in this case. --Kslotte 15:31, 18 February 2010 (UTC)

See here though it is not rendered as far as I can see. --Skippern 10:31, 19 February 2010 (UTC)

Conclusions about opinions

The proposal page has been re-written, since it seems like making the width=narrow obsolete is in favor. Giving some definition for "narrow" hasn't gain much support. --Kslotte 20:52, 18 February 2010 (UTC)

Width rendering

How is the renderers currently taking into consideration about widths? width=narrow and width=some meters? --Kslotte 17:13, 18 February 2010 (UTC)

If width=narrow should be avoided?

What should we do with the tag if the majority want to avoid the usage of width=narrow? Made it obsolete or only make texts that it should be avoided? Can maybe "Keep It Right" catch these? Obsolete or avoid? What is your opinion? --Kslotte 19:12, 18 February 2010 (UTC)

Needs a Bot

If any change such as this is implemented then someone needs to create/find a User:BotItalic text to find and mark or replace the old keys

• Any good coder's/taker's around here?

--Chief Mike 14:32, 19 February 2010 (UTC)

For what do we need a bot? If the tag width=narrow is made obsolete, then Quality Assurance tools (like Keep Right) will give out warnings about it. As time goes by tag entries will be corrected by mappers. --Kslotte 16:49, 19 February 2010 (UTC)

Why I believe both options (narrow / exact measurement) are needed

My reasoning is: Let mappers do the tasks that their brains are suited for. For a human, it's easy to see if a road is narrower than normally expected. But it is hard to guess the width in meters or feet. So in most cases, width=narrow works best for them.

If you apply this reasoning to computers, you get to think about map renderers. They can either apply the default road style, or some special rendering which indicates a narrow road. If they get e.g. width=3m, the renderer must decide whether 3m is narrow or not in the given context. Given exact rules about it makes this an easy task for the computer, but again it's hard to code all these rules.

Anyway, an exact width has practical advantages, for example when a navigation software known the width of the vehicle. I think in places where the width really matters there are road signs like the German Z264 which indicate the width, which makes is easy to map.

To make rendering easier I think even when an exact width is known, it might be good to clearly indicate "narrowness". Of course, you can't write width=2m and width=narrow at the same time, so we'd need yet another solution. --Brian Schimmel 10:28, 23 March 2010 (UTC)

I don't agree, if you need to estimate user estimated_width, not narrow! Lulu-Ann

Key:width edit waring

We have a edit waring going on page Key:width.

User Pieren wants to add the following text: "Note that tagging width=narrow is not recommended by some contributors (by few more voters here: decision). This value can be used by the others who prefer to express what they see instead of estimating bad distance numbers.".

This text should be removed, since consensus have been achieved both by discussion on this page and by voting to make narrow value obsolete. Comments regarding this is wanted. --Kslotte 14:44, 29 March 2010 (UTC)

Kim, you should perhaps contact me before anything else... Anyway, we understand that you are strongly pushing to deprecate this tag. You said "it's a consensus". No, as it was already in the thread you initiated on the ML ([1] and the 29 answers) and again on the 'voting' page you creating, they are a significant number of people who do not want to deprecate this tag. 6 or 7 more voters in favour of deprecating is not significant enough compared to the tens of thousands contributors of OSM. On my dictionnary, "consensus = An opinion or position reached by a group as a whole, general agreement". It's not the case here. I would like to see you spending the same efforts on clarifying the questions raised on the ML thread like a clear definition of the width boundaries. --Pieren 14:58, 29 March 2010 (UTC)

Pieren wrote "Anyway, we understand that you are strongly pushing to deprecate this tag". This is not true; as I originally wanted to leave "narrow" and give it a definition. But since there where much resistance and no support for this I make a decision to go the other way and propose to deprecate it. I myself have taken an objective position in this and asked for opinions. I haven't even voted myself. Everything is according to the processes for proposed features. --Kslotte 15:11, 29 March 2010 (UTC)

Okay. But you have to know that many contributors ignore this process as they consider that vote is done throught the tags usage statistics. It's not my case as I think that votes (I prefere 'opinion poll') is always good to put the arguments on the table. As it was said on the ML, widht=3 (meters) is better that width=narrow which is better than nothing. But I will not insist to preserve your own proposal. --Pieren 15:28, 29 March 2010 (UTC)