[OSM-talk] The future of bugs in OSM

[Matt Amos]

On Thu, Jul 2, 2009 at 1:19 PM, SteveC<steve at asklater.com> wrote:
>
> On 2 Jul 2009, at 09:44, Dair Grant wrote:
>
>> Frederik Ramm wrote:
>>
>>>> And if the user indicates that he just wants to add a PoI,
>>>> redirect him to
>>>> http://ae.osmsurround.org/ so that he can add it directly to the
>>>> database.
>>>>
>>> That's the point I was trying to make - do not hog all the bugs in
>>> one central
>>> place and allow users to do only what you have coded; instead open
>>> this up so
>>> that anybody can hook their app into the user interface to offer
>>> functionality.
>>
>> Is the bugs database really so different in character to the map
>> database
>> though? Provided there was a planet-style dump of the bug database,
>> anyone
>> wishing to build an external system could easily do so.
>
> good point

it's not just a case of linkage to external systems. i see two
different generators of bugs; users and tools. we have to think about
the different requirements of each - users will need a *very* simple
interface for inputting bugs (OSB is a good example), whereas tools
will need/want a more complex one. it's worth putting in a little
effort at the beginning to make sure that we can meet (or evolve to
meet) both sets of needs or the current situation of silo'd data will
continue.

>> It's true that lat/lon/text wouldn't be sufficient for all bugs, but
>> how
>> many would that model work for - 75% or more?
>
> see the tags idea

non-OSMer users will want lat/lon/text to be sufficient, with maybe
optional tags. the tools will want far more information - maybe
inserting an FK id for their own systems, or having a "ping-back" URL
to keep their data in sync. as frederik pointed out, a single lat/lon
may not be enough - some systems may want to link a bug to specific
OSM element(s), or provide a bbox.

> these are all good points, I say lets build the simple bits and iterate.

exactly. let's have a discussion here and on #osm (this evening
european time) to figure out the simplest solution. i've already
started building something and, as soon as it is non-simple enough to
be useful, it'll go into SVN and we can start iterating.

cheers,

matt