[Tagging] tagging of one-way cycle lanes

Marc Gemis marc.gemis at gmail.com
Fri May 11 16:18:05 UTC 2018


On Fri, May 11, 2018 at 5:40 PM, Paul Johnson <baloo at ursamundi.org> wrote:

>
> Why the almost religious doctrine level of resistance to change?  Even the
> Linux kernel rewrites entire subsystems from time to time when a superior
> approach comes around.

typically such a thing is done in 2 phases (e.g. Java language)
announce that the old API is deprecated and offer the new one in
parallel. Then, after a grace period, drop the old API.
This change is well announced, and it is clear which results you might
expect with the old and the new API.

Now apply this to the lanes-tag: Day A it means only "full width". The
next day it means all lanes, but none of the objects with this tag are
updated. Now, people will revisit those places one-by-one and update
the value. But perhaps some mappers did not read the new definition
and keep adding lanes-tag according to the old definition for awhile.

Now try to find out what the meaning is of the lanes tag on any random
object in the DB. Will you be successful ? I hardly doubt so.

There is no problem adding a new "lanes_for_all_vehicles"-tag, as
everyone using that tag knows that they have to count the cycle lanes
and perhaps pedestrian lanes as well.

Changing the meaning of a tag or value is impossible since the data
will not tell you whether it is put in the database according to the
old or the new definition. We have this problem even at this moment
(since you apply another definition than many other mappers), but we
can refer you, new mappers and  data consumers to the wiki page and
say this is how it should be done. We do not wilfully introduce
ambiguity in the interpretation of tags.

regards

m.



More information about the Tagging mailing list