waymarkedtrails / waymarkedtrails-backend

Database backend and rendering styles for waymarkedtrails website.
GNU General Public License v3.0
15 stars 6 forks source link

Support route classification in `network:type` as well as in `network` #16

Open quincylvania opened 3 days ago

quincylvania commented 3 days ago

I recently started giving US trail routes a proper network tag value suitable for filtering and rendering, similar to road network. For example, I've been tagging National Scenic Trails as network=US:NST. I moved the nwn value to network:type, a pattern which already has a few thousand uses. Unsurprisingly, this has broken Waymarked Trails. I'm hoping you'd consider supporting network:type in addition to network for classification, perhaps detecting which is which with regex or a set of known values. I'm happy to open a PR myself if you think this approach makes sense.

lonvia commented 3 days ago

I'm very much in favour of fading out the difficult network tag and can hack the waymarkedtrails code to help with the transition. Your suggested way forward looks feasible. Two comments though:

  1. It would be good to have a bit of a community discussion first. I know that this is not an easy request to make and I'm fully aware that such a discussion is very unlikely to come to a meaningful conclusion. That's fine. The tagging you propose can to a certain extend exist in parallel (although data users will be forced to support both, oh well). However, it would be good if it eventually becomes the world-wide standard. And for that it always helps when mappers from other countries have a chance to look at it and check how it fits for them.
  2. Please, please, do not repeat the mistake of the [lrni][wch]n value pattern. If we are moving to a new tag, then lets move to meaningful words that are usable for any mode of transportation and doesn't end us with semicolons on multimodal routes. local, regional, national is probably fine. international has always had the problem that some people took it to mean "crosses a boundary" (as opposed to "is so long that it covers multiple countries"). I tried to sneak in continental into waymarkedtrails but I can't say I've ever liked the term. Just couldn't come up with anything better.

I can already say that you will get some opposition from some node network users, who have stated in the past that network:type=node_network and network=[lrn][wc]n are orthogonal. I've closed that discussion here in the repo as doesn't make any sense to me but that's just my opinion.

1ec5 commented 2 days ago

Some of the disconnect within the community may be lack of mutual understanding about what constitutes a network in the real world. This forum thread explores the relationship between route markers and networks, and some of the considerations that lead mappers and data consumers to shy away from certain tagging schemes. It isn’t specifically about network=* or network:type=*, but it would benefit from more attention from people who understand how renderers work.

lonvia commented 2 days ago

Shields and network tags are a completely orthogonal discussion. I wouldn't mix them.

wegerj commented 2 days ago

Bear with me. My two cents here. I apologize if I am out of my depth. For example I do not already understand how "Shields" and "Orthogonal" are used in the OSM context. My github account is just a past join event.

I did a bunch of mapping of the CDT. I have section hiked hundreds of miles of the trail. I am a business applications coder of many years. In later years I worked with MS-SQL.

I too found the "Continental/International", "IWN", value problematic. It works for Europe but not really for the US. The PCT, AT and ADT trails and others would be "IWN" trails in Europe.

I was glad WayMarkedTrails supported nested relations as that greatly supported my mapping efforts.