[OSM-talk] Locks

[David Earl]

> -----Original Message-----
> From: 80n [mailto:80n80n at gmail.com]
> Sent: 08 December 2006 19:57
> To: David Earl
> Cc: OSM
> Subject: Re: [OSM-talk] Locks
>
>
> On 12/8/06, David Earl <david at frankieandshadow.com> wrote:
> I recently had occasion to add a couple of locks to the OSM.
>
> Currently the Map_features page says waterway=lock_gate is
> defined on nodes.
> That's fine for the creator (me), but it occurs to me that it
> will be hard
> to render when it comes to be done - assuming it would be
> represented as on
> a OS map for example...
>
> ----<-----<---
>
> - you'd have to determine which of the ways connected to the node
> are river
> or canal (very often footpaths go over lock gates, so there will be other
> things conected to the nodes; and what if there are more than two - if a
> weir section of the river goes off from that node too for example?)
>
> A footpath shouldn't connect to the nodes that make up the canal.
>  That is, unless, the footpath actually goes through the water ;)


You missed the point. Footpaths often use the lock gate as part of the
footpath and the node _is_ the lock gate, not the river. It's not a bridge
in this case.

In any case it's no good saying something _shouldn't_ do something, unless
it is in the spec, as you can be sure it will happen if it isn't. Indeed
that's true even if it is in the spec but not enforced.


> If there is a branching channel then that could come from a very
> closely adjacent node.

True, but same point as above

> It should, imho, be a rendering error to try to draw a lock gate
> on a node that has more than two segments connected to it.  There
> are other similar cases, like gates, where the symbol should be
> marked by a node but drawn at right angles to the path.

Indeed.


> The direction the segments should determine the direction of flow
> of the canal (as it does for rivers) and hence the direction in
> which to draw the lock gates.


(Where does it say that rivers should be in the direction of the flow?)

But my main point is that this is a much harder calculation than based on a
single segment, and is very vulnerable to how the map data is represented in
ways which don't follow assumptions which aren't available to contributors.

I still think representing the lock rather than the gates is a better
representation.

The lock (gates) I added do indeed have the properties you've mentioned,
except possibly the direction of river flow which I'll have to check, but
that's more by accident than design.

David