[OSM-talk] MEP - pipelines

[Lists]

> On Jan 4, 2015, at 21:11, Richard Z. <ricoz.osm at gmail.com> wrote:
> 
> On Sun, Jan 04, 2015 at 09:18:57AM -0200, Lists wrote:
>> May I suggest the following work flow:
>> 
>> 1) Agree upon 2 dates sufficiently spaced i.e. 2 weeks apart
>> 2) First date, add the new tag, leave the old tag in the system. This will not break anything, and from that date data consumers know they can start migrating their renderers, harvesters, or whatever into the new scheme without loss of data integrity
>> 3) Run sanity checks on the data, maybe manually correct slips such as type=sewer -> substance=sewage if needed
>> 4) On Second date remove the unwanted key, objects where the key should remain should have been flagged by now, and data consumers should all be on the new scheme also.
>> 5) Define an interval to re-run the scripts if it is probable that anybody still might use the old scheme.
> 
> it needs more flexible rules and generally the suggested 2 weeks are 
> extremely short.
I was just throwing out a suggestion for timeframe without thinking much about its feasibility, I agree that the timeframe might need to be longer, and that is for me an open topic for discussion. I will not myself take part in any mechanical edit, though I might be affected as a data consumer.

> For example we still have many (not counted) culvert=yes even
> though it is considered obsolete since a long time.
> 
That might be that they haven’t done any mechanical edits, and that old data still are accepted in many data consumers

> In those cases (like here) when it is possible to have the old and 
> new tagging in parallel the transition could be done in two steps 
> long apart.
> 
I agree that it should be a two-step job
> 
>> This should be the general rule for mechanical edits when migrating tagging schemes. Also make sure that editors such as Potlatch2, iD, JOSM and Merkaartor, all have corrected their presets by the second date. If old scheme is stuck in the preset than it is likely that old scheme will continue to be used.
> 
> also to be considered if keepright or similar need adjustments
> 
I was mentioning a few sources I could get from the top of my head, I probably have forgotten a lot of them
> 
> Richard

Aun Johnsen