[OSM-talk] Thoughts about Map Features and Maplint

[David Earl]

On 27/11/2008 00:18, sylvain letuffe wrote:
>> I'm going to muddy waters between two threads that are going on here.
> You are more welcome to do so since your proposition seams a rather very good 
> one.
> 
>> formal part of the API to update it, so you could in principle introduce
>> new tags (with descriptions), as well as translations for existing tags,
>> through the API via an editor (either one built for the purpose, or
>> through JOSM/Potlatch/whatever, or both).
> That would be, by far the BEST way to go imho, what's best to describe the 
> content of a database than a table of the database itself ?
> 
> But this sounds like a major proposal that has to be carefuly though of.
> ( I have no clues to who will make the code)
> But I have right now many fears that comes to my mind. By order of magnitude :


I think I need to write something down that is more concrete than the 
mail threads, so you can all see how it would work.

> 1) Won't that end in a way to enforce possible tags and makes it impossible to 
> tag my own ?

Not as I am suggesting it. What would be required is that you don't 
_casually_ introduce a new tag - you make a deliberate decision, and 
provide a small amount of info about the tag, including a short 
description in at least your own language. (In principle an editor could 
sidestep this by implementing something which says "if tag not found add 
it with an empty description", but that wouldn't be helpful).

> 2) won't that remove my freedom from proposing new tags ?

Anyone can propose anything anywhere :-). And as it doesn't stop you 
implementing it, you are no worse off than now (unless you regard being 
asked to be a bit more informative when you do choose to add a new tag 
as being worse off - personally I think it makes you better off as the 
tag has more chance of being widely adopted if others like it, because 
editors can, in pricniple, get to know about it automatically in their 
presets and provide help for it, and because people then see it they 
would start to offer localisations for it)

> 3) won't that end in an horrible mess of thousand of duplicate tags ?

Not the way I was proposing it - see the other thread. 
(http://lists.openstreetmap.org/pipermail/talk/2008-November/031769.html 
- especially my reply at the end)

I was specifically thinking it should be upward compatible and not stop 
anyone from doing anything they can already do (though it might slightly 
change the way you do it, and require you to make an extra step when you 
want to introduce a new tag, but I'd expect this to happen as part of 
the existing editing - JOSM could prompt you for the info, for example)

> 4) won't that look rather like the output of tagwatch ?

Yes, except less anarchic.

> If those can be adressed correctly I am much much in favor of that, than every 
> editors having their presets, having maplint parsing a wiki page, having a 
> too big map features page.
> 
> How do we start to talk about it ? are some devs on the problem ? is this a 
> virtual proposition for now ?

I've not done any of the API programming, but I've certainly read the 
code and contributed to development in JOSM and the namefinder, so I 
think I can qualify as a "dev" (pretty please). I know what would need 
doing and where, and could in principle write it. But I don't recall any 
of the people who've been doing API0.6 joining in the discussion so far.

But yes it is only something that is crystallizing in my mind after the 
discussions of the last couple of days.

I think the next step is for me to write it up somewhat more formally, 
which I'll do in the next few days.

David