[Tagging] Using route=road to group the segments of a street (was: Re: Reasons for associatedStreet?)

[Nathan Edgars II]

On Fri, Sep 3, 2010 at 11:53 AM, M∡rtin Koppenhoefer
<dieterdreist at gmail.com> wrote:
> 2010/9/3 Anthony <osm at inbox.org>:
>> On Fri, Sep 3, 2010 at 11:42 AM, Nathan Edgars II <neroute2 at gmail.com> wrote:
>>> Split a way and save the change without updating the relation that
>>> contains the way. Voila, broken relation. This happens more often than
>>> one might think; for example downloading relation members in JOSM
>>> doesn't check for and download any relations that contain those
>>> members.
>>
>> Sounds like a bug in JOSM, no?

Yes and no - unless you want it to check for relation membership
whenever a way is split (which makes it impossible to work offline),
there's no way to completely avoid the problem.
>
> I don't understand the problem regarding JOSM. If a way is contained
> in one or more relations and you split it, you get a popup where
> according to the relation-type a default is set, and you are asked to
> modify or confirm. I have seen this for multipolygons, routes and
> turnrestrictions. Can you please specifiy which type of relation you
> refer to (and/or which version of JOSM you are using)?

This only works if JOSM has the relation downloaded. There are many
ways to load a way into JOSM without checking for referrers.

Another way a relation can get screwed up is a truncated save in
Potlatch, again from splitting a way. If the new ways are saved but
the connection dies before the modified relation is saved, the new
half of the way will not be in the relation.

I don't think there's any way to make this foolproof without changing
the way ways and relations are stored in the database. So we should
keep the tags on the ways for purposes of redundancy and repairing
relations when these problems do happen.