[Historic] Historic Digest, Vol 6, Issue 15

[Robert Warren]

Tim,
Christian,

I suggest to use the same approach as start_date [1], default to Gregorian, allow change by separate "j:date, xxx:date" prefix for another calendar.

The examples shown on the webpage use double periods for ranges [2008-08-08..2008-08-24] instead of colon, which might not be a bad idea in case someone starts putting full timestamps in there.

best,
rhw
[1] http://wiki.openstreetmap.org/wiki/Key:start_date


On 2013-01-23, at 1:22 PM, historic-request at openstreetmap.org wrote:

> Message: 2
> Date: Wed, 23 Jan 2013 16:08:31 +0100
> From: Tim Alder <tim.alder at s2002.tu-chemnitz.de>
> To: historic at openstreetmap.org
> Subject: Re: [Historic] Temporal Tagging - old names
> Message-ID: <50FFFCEF.1000609 at s2002.tu-chemnitz.de>
> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
> 
> Hello,
> I like the python-style for the time range definition and this style 
> don't break with OSM-rule of unique keys for each object.
> 
> What we should additionally handle is to work with uncertainties.
> So in my mind there are different kinds of sources:
> *If we found documents with the exact year it's easy:
>   name:[year1:year2]
> *If we find only two historical maps with a streetname and I know when 
> these maps were printed. I know that the street got the name before the 
> first map was printed and had this name also after the second, younger map.
> (Problem is that changes need in the past some years to come to a map, 
> also today it can need a while). In such a case I would propose to use:
>   name:[<year1:>year2]
> *If a street has the name from a person, it's mostly so that the street 
> didn't get the name before the person was born. In our example Emperor 
> Wilhelm I. was born 1797. So I would say:
>   name:[>1797:1936]=Emperor-Wilhelm-Street
> Perhaps somebody comes later with a map that shows that the complete 
> street didn't exist at this time. So I hope we find a system that 
> becomes better and better over time with the help of many users without 
> to be perfect at the first day.
> 
> As standard for time we should reference at ISO 8601, so if people know 
> it exactly they would be able to use YYYY-MM-DD instead of YYYY.
> 
> To make it not to complicated, I hope the gregorian calendar is ok.
> 
> 
> Greetings Tim