On 20/08/19 18:38, Richard Fairhurst wrote:
Kevin Kenny wrote:
There's also something to be said for using the ugly editors to
prove the concept, because at this point, we don't yet know how
to do everything, much less how to make it novice-friendly! The
exception is simple linear routes, and Sarah or I can give you
algorithms - or at least heuristics - for maintaining sort order
on those.
I have an algorithm like that too - it skeletonises dual carriageways and
roundabouts, hops over small jumps, and so on. But that's very different
from the steps to implement in an online editor, which has many more
constraints. (P2 doesn't have access to the full set of JTS/PostGIS tools,
for example!) _If_ the issues can be identified clearly and the realistic
steps to fix them enumerated, then we're getting somewhere.

I do want editors minimally to observe the 'don't break the route'
principle. About 80% of the broken-route problem can be solved
simply by, "when splitting a way, both the pieces become members
of any route relations in which the original way appeared, with the
same role if one is specified, preferably preserving continuity if
either or both endpoints was shared with the neighbouring way
in the relation." At least iD, Meerkartor and JOSM all do that.
As does P2, I believe (I didn't write that bit of code) - iD's code might
actually be based on P2's. That does make me wonder how much of a problem
this is in reality if the four major desktop editors already support it.

I don't know.

I think some separate a way and insert a new roundabout .. the new roundabout 
does not go into the relations.

I have had some that were broken by the additions of turn restrictions. Never 
bothered to find out what they did, just fixed them and got on with life.
Usually they are iD edits, I suspect most people use iD, particularly those 
just starting out so it is no real guide as to potential iD problems.


For what it's worth, I think that the "route editing is complex"
problem partly drives the 'startled warthog' and '1980s throwback'
issues. In my experience, newer and prettier UI's try so very hard
to be pretty and novice-friendly that in many cases, they simply
reach a ceiling of complexity beyond which they can't cope or
become an obstacle to the power user.
Generally I tend to think that a data model that can't be edited with a
simple UI is a bad data model; and that "power users" are a curse on
Wikipedia and rapidly becoming the same in OSM, especially when their main
role is to generate abstruse content as self-gratification but which no-one
will ever actually consume. But that's just me being a grumpy old man too.
:)

That is an expanding club. Generally life membership is granted.


_______________________________________________
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Reply via email to