[OSM-dev] Grumble, grumble

[Matt Amos]

On Mon, May 11, 2009 at 5:17 PM, Stefan de Konink <stefan at konink.de> wrote:
> Shaun McDonald wrote:
>> On 11 May 2009, at 15:13, Dave Stubbs wrote:
>>
>>>>> i don't understand what you're trying to say. changesets have
>>>>> bboxes,
>>>>> yes. they're calculated from the data, not exposed in the URI like
>>>>> XAPI does.
>>>> I start editing Josm/Potlatch/Merkaartor says: for the next 15
>>>> minutes
>>>> bbox lat/long is locked for editing. For properties highway=* etc.
>>>> The
>>>> editor pings back after the period to maintain a lock.
>>>>
>>> That sounds wonderfully scalable.
>>
>> and will work really, really well at mapping parties.
>
> ...and at a current mapping party NOTHING will work because of exactly
> the same issue you see now. Btw at Dutch mapping parties we already bbox
> the area someone is going to do, so if you didn't have any sarcasm, you
> might even have a point.

at a current mapping party everything works just fine. frederik has
already explained how JOSM is going to deal with this and it doesn't
need bbox locking on the DB. i assume richard is planning something
similar for potlatch (?).

bbox locking has its own problems, including easy DDOS, lack of
concurrency, etc... but consider the mapping party example. if there
is a way intersecting two editors' bboxes and they both lock their
bboxes. now, neither of them can edit that way since it intersects the
other's locked bbox.

cheers,

matt