Now that the arguments on both sides have been repeated a couple of times, I'd like to offer my solution; me and some nearby have been using this for some years already.
First, I believe, why the points mentioned are incompatible: There's two ways to look at the keys (not just this key): 1) anything with railway=* is "some sort of railway right now"; the humanitarian map layer seems to consider the key like that, every way with railway=* is rendered equal. If the track is "abandoned", the soil and right to use the land is intact, and new tracks could be laid down relatively easily; not a usable railway, but a big portion of the structure is still there. In this case, a railway=dismantled is internally invalid; it's no longer "some sort of railway right now". 2) things tagged with the key railway are somehow intrisically related to the rail tracks; signalling, water points for steam locomotives etc. The same viewpoint is used sometimes even with the key highway: highway=street_lamp is not a highway, but it was considered so essentially related to the highway, that it would have been possible to just fetch all objects with highway=* to have the important parts of the highway environment. Even barrier=gate's were highway=gate in the beginning. If one uses this viewpoint in all their interpretations, the former course of a railway, even if only verifiable from old documents, is somehow related to the current day rail network, i.e. belonging to the key railway=*. Neither of 1 and 2, above, are always correct. I have some insight on bits of old track in urban environments, so I'll use them as examples. Near me, there's a straight opening in the wood, somewhat elevated from the surroundings. There's no visible path on it, and there could be buildings on it in the future. The rails were removed in 2000, and one might find some remains of the auxiliary structures. Clearly, a railway=abandoned on that section. Where that track used to connect with the present day tracks, a road for buses only was built in its place (in the center!); the old railroad bridge even remains standing as a part of the road. The tracks were actually left behind for several years, and it was changed from disused to abandoned just last summer: the embankments, cuttings and the layout still remains. Near the cemetery, a long straight cycleway across some fields etc. turned out to have been built on a former railbed. Only where it crosses a small stream, one might be able to visually identify the past. None of the other cycleways in the area are that straight, and the orientation of the straight seems out of place; the fact that it was a railway is great knowledge. Elsewhere, there's a long curved cutting in the rocky hillside near the former harbour. The curve turns out to be such because a freight rail track used to run there 60+ years ago; for all I know, the curve is likely to stay in place for decades. In the city center, there's a building with an exceptionally high loading dock, because the building used to be harbour warehouse with a freight track for loading and unloading right where the present day sidewalk is. As long as the building is standing (and it's likely to be protected, if it hasn't been protected already), there are visual signs that there used to be a railroad. moltonel 3x Combo wrote: >railway=abandoned without glancing at the satellite imagery (no, >Also, if an abandoned railway has evolved into something else, then >it's not an abandoned railway anymore. If you add a highway=cycleway The solution: Tags are cheap. I have mentioned the idea in the past, that when any feature is removed because it was destroyed, one could first prepend "was:" to every key, set end_date=*, upload to server, and only then delete the object from the database. That way it would be at least stored somewhere that the object was removed because it no longer exists. Hidden in the full history dump, but it didn't vanish without a trace. Some have used the prefix "historic:", but I prefer "was:" because it's shorter, clearly indicates it no longer is that, and is almost at the end of the alphabetic sort order. Extending this, when there's nothing left of the rail track, change railway=rail (or railway=abandoned/disused) into was:railway=rail (or was:railway=abandoned etc.), set end_date if you know it. Now, it doesn't anymore try to claim it's a "some sort of railway right now" - it's not tagged railway=* anymore - but it conveys the past, no matter whether the relevant parts of the ways are reused for footways or whatever, or whether the ways run through a void. If the area gets extensive reuse in some other form, and the ways get in the way of editing, the next editor might remove them. If not, they don't then do no harm. Elsewhere near the center, a cycleway was built in a deep trench, right where the tracks used to run; the existence of the trench can be explained with one or two simple tags on the cycleway: * was:railway=rail * end_date:railway=2009 * (was:service=spur) Richard Fairhurst wrote: >attitudes that have damaged the Wikipedia community so. Please >let's not adopt deletionism as well. IMO even if the was:railway=rail ways crossing a present day building sometime get deleted, the fact that the old features shared the space and the database with the present day objects is beneficial if one later wishes to reassemble the features from the full history dump. I dare to presume that in the end even the former railroad ways crossing new developments will not make editing so much harder that users would usually even notice them; the level of detail in inhabited areas is increasing rapidly that editors need to zoom in really close anyway, so the past railways with sparse nodes hardly ever get in the way. -- alv _______________________________________________ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
