[OSM-talk] Proposal for a map-bug tracker (Openstreetbugs)

[Christoph Böhme]

Hi,

Xav <x at nainwak.com> schrieb:
> With my experience in developing OSB, I would say that Christoph just 
> resumed it quite right : the server side software is a piece of cake
> and should propose a simple API to insert/edit/delete and view the
> data (JSON, RSS, GPX).

That sounds good. I will see at the weekend if it really is a piece of
cake. Would it be possible to reuse and extend the client-side code from
osb for a web-based client-side interface?

> Because everybody has its own idea of what should be specified in the 
> data (bug status, email, classification, age of john's mother), why
> not to copy OSM : tags.
>
> Think of it...

I thought about using a general tag scheme too, but I think its not a
good solution for a bugtracker. Bug reports are mostly free-form text
already and contain only structured information to remind people to
supply certain bits of information and to handle processing of the bug
reports. So, I do not think bug reporters will ever feel the desire
add tags to their bug reports. In fact, it would probably confuse most
people. Developers of user interfaces for the bug tracker might however
want to have more structured information. But this is probably only a
small group of people who can decide which information a bug report
should contain. Especially it would not help anyone if bug reports
contain different information depending which user interface was used
to add them to the database. Just imagine the situation where a user
adds a bug through the web interface and a mapper requests the bugs with
JOSM. Both pieces of software need to use the same model of
information in the report and the same concept of how to process the
bug report.
Additionally, I think defining a bug report format is not like defining
a database structure to describe the whole world but more like finding
one for describing a residential street. Implementing a general tag
scheme just postpones the decision of what to put in a bug report in my
opinion.

> The server side already exists : it's basically OSM database without 
> ways, with a guest account, and accepting long string values (the
> text users could add).

That is really attractive. The only problems I can see here (apart from
that we still should try to define a  bug report format) are that
annotations to a bug report like comments, images and attachments
cannot be stored in the osm database (as far as I know). Another
question is how the introduction of changesets in the api 0.6 affects
very small edits. Creating a changeset for every bug which is added to
the database might turn out to be very inefficient memorywise.

> A lot of clients already exists : JOSM, Potlatch, Mapnik, trillions
> of scripts, etc.

But they still need interfaces to handle the bug reports in a user
friendly way.

Christoph