[Imports] Fwd: Postboxes in Czechia, updating details as import

[majka]

Sorry, didn't notice I was responding privately.

---------- Forwarded message ---------
From: majka <majka.zem+osm+imports at gmail.com>
Date: Sat, 13 Oct 2018 at 16:07
Subject: Re: [Imports] Postboxes in Czechia, updating details as import
To: <matkoniecz at tutanota.com>




On Sat, 13 Oct 2018 at 11:17, Mateusz Konieczny <matkoniecz at tutanota.com>
wrote:

> 12. Oct 2018 13:11 by majka.zem+osm+imports at gmail.com:
>
> The proposed import will be on very conservative side. The nodes will be
> re-tagged only, no positions will be moved, no nodes will be deleted, just
> re-tagged as "disused:amenity".
>
> So there are some postboxes that  are gone according to the official
> source.
>
First, let me say we are talking about edge case in scope of this
import/update. Unless the postboxes of the post depot have all coordinates
in the data souce, no deletion or even retagging will be done, because in
this case I am presuming that it might be a postbox where no coordinates
were given, the address was not precise enough and I could get a false
positive for this reason. More than half of the post depots is currently
out for this reason and I am taking it out completely. For such depots,
I'll be doing only update of the collection times and nothing more

> In that case disused:amenity=* is about as incorrect as amenity=*
>
IMHO, this is not true. The amenity is definitively wrong at the exact
point of time we are talking about.
The postbox is not in the monthly data, and this means no letters are
collected from there. This is the only completely true interpretation.

The gueswork starts after the previous statement and I am simply going for
the import with the interpretation that will cover 9 out of 10 cases:
According to our experience going back almost two years, 9 out of 10 will
be simply temporary out of order, because there is a construction site
going around it, most often on the road/sidewalk/nearby building. You can
still find the postbox on the ground but you cannot use it to send letters,
you simply cannot reach it or the post office workers cannot get to it. In
the most cases, you will find it back in order in the next 3 months, not
many constructions sites go longer than this at exactly the same place. The
construction workers might still be there, but the site wouldn't be
blocking the postbox any more in most cases.
The remaining "missing" postboxes are indeed gone in the meaning there is
nothing left behind on the ground.

> The final delete should come a year after the postbox was last found in
> the source data
>
> Is this delete planned to be done without resurvey?
>
Not always. In the most cases, somebody will delete the postbox after
resurvey:
This is another reason for using the disused:amenity and fixme tag. This
one says in the local language something like ...the post office is
currently not operating this postbox, please go check it. If there is
active mapper around, they will do it. This means places like big cities
and their surrounds are mostly covered.

But some postboxes will be deleted without resurvey, if no active mapper
involved in this task will visit this location in the mentioned year. They
will simply "age" into the automatic delete.
This will be done separately from the "normal" update - I would be simply
using overpass turbo to find all the "amenity:disused=post_box"
"operator='Česká pošta, s.p." and check the source=CP:YYYYMM date from a
year back (meaning currently it would be source:CP:201710). But we are
talking about situation which would for the first time occur in October
2019, not any earlier. I don't expect there will be in any given month more
than 10 cases for the whole Czechia, if something big doesn't happen with
the Czech Post. But there are places even in Czechia, where no mapper was
active in years, not even one on vacation. I will look at every node
indivudually to see, if a note/comment is there opting out of the delete.

And now the reasoning:
If the postbox is not in the data source, the postbox is simply not an
"amenity=postbox" - you cannot use it to send letters.
Direct delete is not suggested because of the previous point (you cannot
say for sure if the postbox is simply out-of-order but there; or if it was
completely removed), but leaving it in the data indefinitelly is wrong as
well. The year should give enough buffer to catch the reactivated ones. If
the presumed construction site is going longer then a year, it would
probably mean it is a big one, the original postbox is not there any more,
a new one was put somewhere nearby and you need survey the location because
of it and all the changes the construction site brought.

>  Is there a wiki page documenting this import?
>
Yes, a very brief one, mentioned in the initial post,
https://wiki.openstreetmap.org/wiki/Automated_edits/majkaz_imports/

I can adapt it if there is a need for it, but most of the discussion
went on the Czech talk list. There was an evolution of the work and
the way we are working with the data source, not all what is mentioned
and spoken about does really concern this proposed import. This is
something what simply grew out of it in the last months, because we
are seeing the need to update our own data from a year ago or even a
month ago, even if these were based principally on the same data
source and we were personally doing survey for it. The same is true
for the documentation in Czech, only a very small part of it concerns
this import.

Majka
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/imports/attachments/20181013/65d2d1f2/attachment.html>