narrow streets in old cities

thanks for the hint, in fact I am also editing Safi (أسفي) area (http://www.openstreetmap.org/?lat=32.29875&lon=-9.2407&zoom=17&layers=O) and I thought I would make some experiments in the old medina, by checking which representation I like most in both mapnik and openstreet and I think I am going to tag either as residential or service/alley (discarding use of pedestrian and living_street). did not think of footways nor path, I will test those as well.

don't go too much for the rendering appearance, this might also change. Try to represent the actual situation (legal access rights, physical attributes like surface and width).

ok, thanks, I will do so and let's see what happens. I noticed a previous mapper had used the highway=service, which did not match the usage but provided a better rendering. when you say "the mapnik layer on the front page", have you got a link to the sources?

This is what is said since 3 years to mappers. But it is not supported by any tool since then which means that is not so simplistic. And most of the mappers just want to say "'it's narrower", they don't want to measure the width with a tape or even put rough estimates. This solution is simply too fastidious for the average contributors but this is not admitted by some of us. It is also not saying 'it is narrower' since you don't indicate what is a normal width. So you force people to put the 'width' everywhere otherwise you don't have any reference and rendering will be wrong anyway.

@Pieren: do you mean that Gnonthgol's answer is "one" and not "the" way to go (most mappers go)? In my opinion one should not use a wrong classification just to get better rendering! on the other hand, while mapping one wants the rendering to match the observations... have you got links to places where this has been discussed?

@Pieren But it says how wide the road is in reality and if your current vehicle can fit in. That is more important than relative width of streets.
And for the record specifying default tag's values based on areas (eg. administrative boundaries) with inheritance and possible change on lower levels is a feature that I personally miss and that would make the mapping easier/more consistent.

I would not use "highway=residential" for a road that is not passable by at least one car. Adding a width and expecting the renderer and router to fix this is not a good idea. I have been to Morocco and I would tag the "streets" with 1,50 width as service and service=alley like suggested in the top answer.

If cars are allowed on those streets, then they should not be mapped as footways or paths.

This is, of course, highly situation-dependent but a general advice to tag these ways as footpath or track just seems wrong.

the streets I'm describing are paved (not tracks) and are used by (motor)bike so not footpaths. it's streets that locals will advise you not to use, as they "don't lead anywhere". but I want to describe them, not to advice about whether or not to use them.

what about using highway=living_street in combination with some access tags in these cases? motorcar=no for example, if it is physically not possible to drive a car there.

I suppose, these small streets in Morocco are exactly this: living streets, even if they do not have this sign with playing children. I guess, people sit outside in front of their houses, drinking tea and smoking waterpipe, playing backgammon, laughing about the stupid tourist that got stuck with his Rav-4, because the navi said so :D

Also, a maxwidth=n can work wonders here...

@moszkva ter: maxwidth is a legal tag, you should not use it for physical characteristics (use width instead). They are also not living_streets because they are too small (a living_street by OSM definition is a residential street with special regulations like lower speed limits and special parking restrictions)