I’ve been looking at how to quality check the Romanian E8 hiking route relation that I recently re-mapped, as well as how to map the Sammiana Macedonian Endless Trail that I’m working on now. So far these have been mapped as one (E8) or more (Sammiana) route relations that include several sections with roles tagged as alternative, excursion and approach.
Quality assurance tools such as at knooppuntnet.nl and Waymarked Trails focus on the number of segments that a route relation contains, which should ideally be 1. Of course if there are more than 1 segments because there are gaps between segments, this is a problem, but I am wondering if it is also a problem if the number is >1 because there are sections tagged with other roles…
This post suggests that “to make life easier”, the segments with additional tagged roles can be placed into separate route relations, so that the main trail is one segment. These can then be grouped together in a super-relation, increasing the level of relation nesting by 1.
I have started applying this suggestion to the Sammiana I4 route, but started to doubt if what I am doing is right. It may make the life of the quality assurance tools easier, but it doesn’t make the life of the mapper easier. I feel as if I am “tagging for the quality assurance tool”.
What do you think? Should segments of (hiking) route relations with other roles (alternatives & branches) be in separate route relations to make quality assurance easier, or should they be in the main route relation?
I’d say this depends on the length of the relation and excursions. If the route is long enough to be split into sections anyway and the alternative/excursion/approach is relatively long, then put it in its own subrelation, otherwise just add the ways with a role to the relations. Basically, I’d make that decision based on what is the most easy to handle for mapper. QA tools really should ignore the members with roles when checking for linearity. (Easy for me to say, of course, as waymarkedtrails has just received an update to handle roles properly.)
Looking at the Romanian E8, the alternative seem relatively short, so I’d just leave them as ways with roles. (btw it looks like the only issue with the route right now is a small section at the end where you have mixed up the forward/backward roles. They need to be relative to the way direction not route direction.)
… and this is a lifesaver! It is definitely very welcome to have a QA tool that follows clear rules. To be nitty-gritty, there remains an ambiguity on routes that have two (or more) main branches, when none of which qualifies for the ‘alternative’ role.
… and this is an issue when QA tools are opinionated, have bugs or diverge between themselves. But when the QA tool reflects consensual good practices, it’s just a pleasure.
There also are cases when routes are nested for other reasons: multiple sub-relations for stages of just for reducing relation size, factoring of route segments between several published routes. In those cases it may be useful to have alternatives managed as relations, because the published routes may have different opinions about alternatives. For instance, Eurovelo routes reuse French national routes without their alternatives.
In all honesty I have not yet found the perfect architecture pattern for these situations.