[OSM-talk] Undiscussed (?) edits removing lesser-used highway=* tags

Andy Townsend ajt1047 at gmail.com
Tue Nov 10 12:50:24 UTC 2015


On 10/11/2015 12:15, GerdP wrote:
> Now what is meant with oneway=yes;no (or no;yes) ? 

(at the risk of stating the obvious) that's likely to be a merged way, 
where a new user didn't spot a difference in a key that they weren't 
looking at before merging two ways.  Just look at the changeset that 
introduced the key, look at the extra nodes added to the way in that 
changeset, and look at what ways were deleted in that changeset that 
those nodes were previously part of.

It's a great opportunity to have "the conversation"* with newish users 
about OSM's data model, and how as well as what they can see to the left 
in iD there's a whole universe of "all tags" and "relations" hiding below.

However, it makes no sense to force people to have to understand this 
stuff before adding e.g. a POI in iD or a simple mobile editor.  We want 
local users with "local area" knowledge and not non-local users with 
"GIS tool" knowledge.

> Is that an error or did someone try to express the same as others do 
> with oneway=reversible or oneway=alternate or oneway=alternating ? Do 
> we have to allow users to invent more oneway tags like oneway=2, 
> oneway="Marsh Lane" without telling him that this is likely to be an 
> error ? 

Bluntly, I'd say "yes".  Better that someone adds something that is 
"wrong" that they can correct later (or another mapper can) than they 
don't map at all.  It really doesn't matter that there's some data in 
OSM that "makes no sense" that data consumers have to ignore because:

<loud>
There will always be "invalid" data in any large dataset, and data 
consumers always need to detect and discard that.
</loud>

It's 2015 - are we still writing stuff that assumes valid data? "xkcd 
327" was years ago (and it was an old joke then).

> Will this be frustrating for any mapper? I doubt that.

We need to think a little beyond people who know about "nodes", "ways" 
and "relations" here.  Anything that says "you can't do that because" 
had better be really clear about what the problem is (and not use words 
like "tag" or refer to OSM's wiki) otherwise this _will_ be frustrating, 
and will stop people contributing.

Cheers,

Andy (SomeoneElse)

* which you've been doing a lot with "odd key values", so thanks again 
for that.



More information about the talk mailing list