[OSM-talk] Potlatch and the destruction of good work (N6 Ireland)

[Richard Fairhurst]

David Earl wrote:

> I think that would be great - I thought about suggesting it, until I
> thought about what it would involve to implement. Because changes are
> committed straight away, Potlatch presumably relies on the current state
> of the database, so it would have to keep a separate record of changes
> made in order to reflect any changes made so far to override what the
> database says. Basically it would need nearly everything a "commit"
> would need I think.
>
> I say this based only on what I think I'd need to do if I was
> implementing that. I have no knowledge about exactly how it works inside.

Actually, I think you could do a sandbox Potlatch pretty easily, just  
by temporarily disabling the 'putway' (i.e. save to database) call.  
Potlatch keeps ways in memory anyway (up to a limit of 500, to keep  
the memory footprint reasonable).

Should just be a matter of inserting a new line 544 in  
http://trac.openstreetmap.org/browser/applications/editors/potlatch/potlatch.cgi:

    if (_root.sandbox) { return; }

plus, obviously, some UI to enter sandbox mode.

It's an interesting idea, may be worth implementing at the same time  
as the Potlatch docs.

cheers
Richard