[Potlatch-dev] P2 - what's left to do?

[Andy Allan]

On Mon, May 31, 2010 at 10:53 PM, Richard Fairhurst
<richard at systemed.net> wrote:
> Ævar Arnfjörð Bjarmason wrote:
>> Before it can be deployed we presumably need alternate editor support
>> in the rails_port so that users can choose and switch between the two.
>
> I don't think P2 is ready for deployment on osm.org yet, even as a
> second-choice editor after P1. It needs a lot of work on the presets,
> and some help pages, before that's feasible.

Quite.

> It also, I guess, needs
> support for token auth.

Not necessarily, although asking people to sign in to approve an oauth
token for an app served from osm.org isn't the smoothest workflow. But
the basic need would be just a static page with the .swf

> Right now, the question is "what's stopping someone putting P2 up on a
> webpage and letting people edit OSM with it?".

Testing. For example, http://trac.openstreetmap.org/changeset/21504
shows that nobody had ever tested saving new additions before this
weekend, since trying to would crash the app! So that suggests that
any testing done so far has been changing tags or moving nodes around,
but not much more than that.

When I was testing previously on stuff more complex than just the test
data on api06.dev it only took me 15 minutes to find a number of UI
bugs (see http://trac.openstreetmap.org/changeset/21111 ). I suspect
the same will be found with testing the data model. For example,
pressing split way then undo moves all the data back to where it was,
but leaves the way marked as "dirty" so the version gets incremented
(with no changes) when you press save (see
http://trac.openstreetmap.org/changeset/21507 ).

So we should try to systematically test as much as we can think of -
make changes, press save, reload the data (and/or check the changeset
via the browser) to make sure that everything works. And do the same
for pressing undo, and then do the same for pressing undo and redo. I
don't think we'll catch every bug but I suspect nobody's tried
catching any bugs yet :-)

The second thing is the yahoo layer overzoom bug, which is very
confusing. Richard, you fixed it for potlatch in
http://trac.openstreetmap.org/changeset/6769 with "ymap2.swf", so
maybe you could do the same again?

Cheers,
Andy