[OSM-talk] Potlatch and the evil semicolons
Richard Fairhurst
richard at systemeD.net
Tue May 27 11:59:29 BST 2008
Dermot McNally wrote:
> It's easy to see why this won't appear on the map. Investigation
> confirms that Potlatch (at least a few weeks ago when I tested this)
> does this when you merge two ways with conflicting tags. This feels
> like dangerous behaviour, a fact backed up by the fact that I
> personally have had to fix at least 5 cases of roads eaten by this
> effect. Is there any reason Potlatch couldn't do what JOSM does, and
> require the mapper to commit to one tag or other in the event of a
> conflict?
Well, I really, really dislike UIs that pop up an alert saying "You've
just done something wrong which you don't understand. It needs fixing.
Would you like to do something (a) you don't understand, or something
else (b) you don't understand either?". Besides, modal dialogues like
that break up your workflow - you'd be perfectly justified in wanting
to go back later to fix it, for example.
So I'm a bit loth to do it the JOSM way (hey, diversity is good). And
Potlatch does, of course, recolour the way based on certain tags, so
if you merge a highway=trunk (green) and highway=primary (red) it'll
become grey, which is a visual indication of what's changed.
But what could work, I think, is putting a little indicator (perhaps
an unhappy face, or an alert triangle, or something) in the tag to
show that there's an error in it. Probably not quite as authoritarian
as Maplint's not-in-map-features, but something to flag up common
errors like this - and maybe there are more we should be looking for.
Feel free to start a trac ticket so I don't forget. :)
FWIW, I find if you get a bunch of errors like this in one area, it's
usually one mapper rather than necessarily a systemic problem - might
be worth contacting him.
cheers
Richard
More information about the talk
mailing list