[Merkaartor] [OpenStreetMap] #1119: Ability to specify icon size in a style

[bvh]

On Thu, Aug 21, 2008 at 10:49:42AM +0200, Chris Browet wrote:
> > Fixed size specified in pixels. This sets the minimum.
> > Real length on the ground when drawing the icon. This sets the
> > proportional size.
> >
> > For example, for a mcdonalds you could set Fixed = 4x4 and
> > Proportional=30mx30m. Since you can reasonably assume a macdonalds to
> > have that kind of a floorspace in reality on the ground there is nothing
> > goin to interfere with that icon.
> Mmm... Not too sure about that one. Do we really want the icon be the size
> of the building?
> Often, as in e.g. churches, we have an area for the building + a node for
> the POI?
> Additionally, in this instance, I map the POI at the actual entrance of the
> building rather than in its middle (I don't know about the others). That
> would produce undesirable effects where the icon would overlap the adjacent
> roads.
> But anyway, it's the same idea, just the stylers would have to be carefull.
> OTOH, if we introduce the "proportionality" explicitly in meters, we should
> do the same for roads as well.

Thing is : roads have an implicit size already due to the existance of
the width=... tag (and in the absence of that tag, the width is
heuristically determined from the highway type)

POI don't yet have that. That is why I think we need to have the stylers
set the size explicitly. And yes, they should be carefull about those
sizes. For example setting a size=100mx100m on supermarkets won't be the
sensible thing to do, even if a supermarket in general has that size.

In my opinion size should probably be interpreted as : the area where
there is probably not a lot of interest around this POI.

What would the alternative be in your opinion, besides settings the size
in meters for roads in the style (which in my opinion is a step
backwards)

> The problem with shorter release cycle is that  we should each time allow
> some weeks of stabilization and bug fixing. 2 weeks stabilization per month
> is hardly feasible.

Well, the idea would be that a shorter release cycle would need less
stabilization?

> I was thinking about a system of weekly "nighties" or something like that.
> That would be "alpha" releases, without warranty of stability, but allowing
> the non-developers (basically windows and mac) users to use/test the latest
> features...

Do we really have users that would do that? Aren't most 'bleeding
edgers' more likely to use linux?

cu bart