[Tagging] noexit=yes on ways ?

[André Pirard]

On 2014-04-10 16:59, Pieren wrote :
> On Thu, Apr 10, 2014 at 4:26 PM, Florian Schäfer <florian at schaeferban.de> wrote:
>
>> True, in the most cases it is not important. But sometimes it is, for
>> example in my T-deadend example, mentioned on this ML.
>> All cases can be covered by the tagging on nodes. The tagging on ways can't
>> cover all cases (deadends with more than one end) and causes problems for
>> example when splitting ways.
> I also explained why the mentionned examples are not a problem. I'm
> ready to change my opinion if I get valid arguments...
No problem? ...
>> Therefore I'd recommend in the Wiki to use it always on nodes to avoid the
>> problems with way-tagging.
> But we don't have "problems" with the tag on the way !
No problem?

Please allow me a little bit of thinking.
If you've got a dead end at one end of a way, and, required or not, you
tag noexit=yes on the way, then QA agents would not detect a "way end
close to another way" at the other end of the way.
This may lead to the funny situation where a way segment has two dead
ends, one declared and the other one a totally overlooked "way end close
to another way" real error [e.g. created later].

> It's true that the wiki has to document the best practices but it
> should not fordid practices that are not wrong, harmfull, unclear or
> ambiguous ! I regret the time when people worked with a more open mind
> in this project.
And if you pardon my thinking again, what I'm saying above is assuming
that the QA agents recognize noexit=yes when it's put on a way.  Did you
check that they do?
I've no more time to loose with this and other QAs, but I've made the
test with JOSM and it does not.
So, the open minded game we're playing is in fact putting noexit=yes
where probably no QA agents recognize them.

Could you please restore my updates conforming to the consensus of
December 2013?

André.



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tagging/attachments/20140410/672976f9/attachment.html>