[OSM-talk] What is an ID?

Lester Caine lester at lsces.co.uk
Tue Aug 2 18:43:47 BST 2011


Richard Mann wrote:
> I think he was joking about reusing ids (he was illustrating the point
> with the sort of daft temporary fix that someone might do ... if they
> were daft)

Well someone has set up the system to reuse them !!!!
http://www.openstreetmap.org/browse/node/1/history
THAT is a joke :(
PLEASE can someone disable that corruption before it goes too far?

> On Tue, Aug 2, 2011 at 5:35 PM, Lester Caine<lester at lsces.co.uk>  wrote:
>> Rather than comment on the id stability thread, I though it better to turn
>> the question around ...
>>
>> In my own data trail, an ID *IS* unique, and is directly linked to the piece
>> of data it relates to. If I change the data then the version number goes up.
>> If I delete the data, then the ID number remains since it identifies the
>> HISTORY relating to that item. The idea that someone is going to 'reuse'
>> id's because they are no being used raises alarm bells. I thought we had
>> history relating to every element nowadays, so how do you know what the
>> history relates to if you reuse the ID? Also the idea of 'renumbering'
>> everything is equally alarming, given that every old changeset would have to
>> be updated?
>>
>> If we are running out of ID's, then a switch to 64 bit is an urgent
>> requirement simply to prevent the existing data and history trail from being
>> corrupted.
>>
>> Now having said all that ... USING the internal ID for external purposes,
>> while potentially practical, is not the most sensible way forward as has
>> been highlighted in several ways. I'm probably in agreement on a 'link
>> server' approach, but I think that there are a number of areas where
>> additional 'ID's need to be stored in the database and managed in parallel.
>> Postcodes, bus/tramstop id's, Airport codes, telephone area codes, and so
>> on. We had a little discussion on parallel databases, and that slots in with
>> the 'link server', but I think that in order to make ANY parallel database
>> system work, then the core ID's have to be stable and consistent, even if
>> they do return 'deleted' ... at which point on needs to be able to ask 'why'
>> and possibly roll back a change that should not have happened? If a item has
>> history then it's ID can't be reused.
>>
>> http://www.openstreetmap.org/browse/node/<xxx>/history simply has to be
>> consistent? Then what may be missing IS a tag for 'split_from' or
>> 'merged_with' but the linked ID's must also be consistent?

-- 
Lester Caine - G8HFL
-----------------------------
Contact - http://lsces.co.uk/wiki/?page=contact
L.S.Caine Electronic Services - http://lsces.co.uk
EnquirySolve - http://enquirysolve.com/
Model Engineers Digital Workshop - http://medw.co.uk//
Firebird - http://www.firebirdsql.org/index.php



More information about the talk mailing list