[OSM-talk] The future of bugs in OSM

Andy Robinson (blackadder-lists) ajrlists at googlemail.com
Fri Jul 17 17:24:50 BST 2009


Eugene Alvin Villar [mailto:seav80 at gmail.com] wrote:
>Sent: 17 July 2009 4:56 PM
>To: Andy Robinson (blackadder-lists)
>Cc: SteveC; Frederik Ramm; Talk Openstreetmap
>Subject: Re: [OSM-talk] The future of bugs in OSM
>
>
>
>On Fri, Jul 17, 2009 at 9:05 PM, Andy Robinson (blackadder-lists)
><ajrlists at googlemail.com> wrote:
>
>
>	Any reason why we don't just put the bugs in osm. They could be
>nodes, ways
>	or polygons and just have a suitable bug=description key value pair
>plus any
>	other tags need (date opened/closed etc). I accept that the editors
>would
>	need to handle the data a little differently and their might be a
>need to
>	track closed (visible=false or perhaps a new visible=closed). Plus
>the API
>	would need to able to deliver just the bug data for the API rather
>than the
>	whole dataset for the bbox.
>
>	It seems daft to me to go reinventing everything when we have all
the
>tolls
>	already.
>
>
>
>-1
>
>I see bugs data as similar to the GPX traces. It should be on a separate
>layer with separate database tables, etc. I agree that we don't need to
>reinvent everything (we can reuse components of the OSM stack) but as Shaun
>said, bug data should be separate from the real data. Existing tools should
>not be retrofitted to ignore such bug info.
>


Very happy to have these recent responses to my question. I was really only
stirring the honey pot a little. What appears to fall out for me is that:

1. A separate database is desirable.
2. We could reuse a lot of the stack, including the tag structure.

I suspect however that those who have experience of software bug track might
have a different perspective. Am I correct? Which is the simplest approach,
since it's the simplest working method that's likely to be what will sit
best with the rest of the project.

So what else is important?

Are points sufficient for marking bugs? I'd say not. I'd like to be able to
set a polygon as well and possibly even a line. No different from a
node/way/poly in geographical terms.

How can I link a bug to an object in the main db? I'll want to do that too,
but not always. Sometimes I'll want to effectively make a feature request (a
bug for an unmapped object/area)

Cheers

Andy





More information about the talk mailing list