The criteria for the classification of highways is their importance for the road grid and not their physical characteristics which are described by tags like surface, smoothness, width, lanes, oneway. Along a specific highway the physical characteristics might change many times the classfication is the same.

Changing the classification when the physical characteristics change would mean for example to classify the sections of a tertiary highway in villages where there are dual carriage ways with physically separated lanes to trunk. In mapnik the color would change from yellow to green which I think would be rather strange and I’ve never seen this on any map.

We had similar discussions already several times, e.g.


I try to avoid futile discussions on that primary/secondary/… classification, just follow the wiki. If the decisive tags are set, a rendering rule should be able to get the road shown in appropriate style. Make sure that both carriageways are drawn as separate ways, with oneway=yes, lanes=2, dual_carriage=yes,shoulder=yes etc. tags. A rendering rule might then say that a road with the combination of oneway=yes, lanes=2 will be treated as a trunk road.
It’s feasable, at least with mkgmap for Garmin maps, as I proofed in a previous message.

Indeed I misunderstood. I thought you proposed upgrading just a section of 344. But it’s the whole 344 and the 344 is rather long. It’s a shortcut to 3, starting and ending at the 3, no gap in the grid. If it’s a dual carriage way along the whole course similar to 3 and may be the signs in Chonburi to Chantaburi direct to the 344 I also think classifying it as trunk is a good idea. It’s similar to the 36 and 32.

But I think it’s definitely a bad idea to classify sections of the same highway differently as it’s done here for the 3375, the 3245 and others.

And it’s not okay to classify highways which don’t have a reference number or where it is (yet) unknown higher than unclassified. See OSM Inpector at this area.



Actually roads with unknown classification should be tagged as highway=road, until a proper survey can be done. In my book highway=unclassified should be reserved for roads which are definitely not maintained by the Department of Highways or Department of Rural Roads.

If a road is known to be a major highway, tagging it as such and with fixme=Survey needed to confirm classification should be okay.

Regarding your specific example, though, that highway is part of 3191 according to this page and this map from the local DOH websites, so it’s at least tertiary according to the current scheme.


Imho we shouldn’t do that due to the situation in Thailand:

  • Almost all secondary and higher highways in Thailand are already mapped as such. But there are a lot of tertiary and unclassified highways left.

  • Often signs are hidden by vegetation or other signs and difficult to spot even when you follow the whole road.

  • Much is mapped just remotely based on Bing without survey or local knowledge.

  • There’s a lot partially mapped or completely unmapped stuff.

Using "highway=road" and fixme in this situation would clutter the map with "roads" and fixme tags. And I'm afraid they would be there for a long time because there are far too few local mappers. And the few might prefer to map unmapped areas and not to fix incomplete work. I don't expect that this situation will change soon. Thus
  • I tag a highway unclassified if it’s clear from a survey or Bing that it’s unclassified or higher but I don’t have a highway number. It’s an added value to see that the highway is no path, track or residential. However I don’t see any reason or added value to tag a fixme=classification.

  • In general I almost don’t use fixme anymore. If something is important to me I map it instead of adding a fixme tag and hoping someone else will do it for me as I expect the chance that this would happen is close to zero. To give an example: I don’t add a fixme=continue tag at the end of a track when I map it just up to a forest temple.