[Talk-transit] catching up

Thomas Wood grand.edgemaster at gmail.com
Tue Jan 13 16:42:19 GMT 2009


2009/1/13 Brian Quinion <openstreetmap at brian.quinion.co.uk>:
>>> I'm wondering if the first step might be to import the data to a
>>> seperate database so that we can render an overlay on to the existing
>>> OSM data without populating the main database.  We could then get the
>>> overlay as correct as possible and give other editors the ability to
>>> double check it before it became final.
>
>> As Peter noted, we can make several passes over the data, each time we
>> generate an OSM file, we can generate an overlay
>> image/tileset/something fairly easily to do a general check of the
>> data.
>
> Maybe I'm miss-understanding what you mean by passes.  I was under the
> impression that you mean doing an initial import, then an update after
> seeing the result, another update, etc...
>
> So instead you are proposing to initially generate a seperate osm file
> which won't be imported into osm until it is the finished result?
> Have I understood correctly this time?
>
> BTW - this includes the implicit offer to help generate such an
> overlay for checking - not just that you should do more work!
>
> --
>  Brian
>

Yes, this is my interpretation of what Peter means by passes.
The other option is to gradually bring more data into the live OSM
database, but would probably prove more tricky in the longrun.
I think data checking is one thing we're going to need to consider
once a first pass has been done. I'm guessing that at a minimum we
should check a sample of each separate region that has provided data
to NaPTAN to check for quirks such as the one we identified with the
Identifier field.

-- 
Regards,
Thomas Wood
(Edgemaster)




More information about the Talk-transit mailing list