[OSM-talk] Validator tags

[Patrick Kilian]

Hi,

>> From my point of view internal=noname and noname=yes both tell me that
>> the street has no name. validate:residential-without-name=ignore tell's
>> the validator not to highlight the fact that there is no name.
> Then we have the same definition.
Good.

>> But I want a _consistent_ way to do that for _all_ of the different
>> checks. 
> 100 percent with you there, that's probably why we came both of us to :
> validate: and internal:
Yeah. Seems like that idea was somehow in the air.


>> Therefore there is a validate:residential-without-name=ignore 
>> too, analogous to validate:bridge-or-tunnel-without-layer=ignore.
> Mmmm ? I see a bit of your point but unclear. Do you mean that you want a 
> general tag system that would, for any tag that you still don't know, trigger 
> an alert  or stop an alert ?
> is your "-without-layer" "-without-name" a key word in your programming that 
> is re-used in many place ? 
> Does if I tag validate:peak-without-name=ignore will stop an alert in maplint 
> without you doing anything ?
No. I just implemented one switch per test in maplint. The test which
are currenty defined can be seen at
http://trac.openstreetmap.org/browser/applications/utils/maplint/tests


>> I'm not believing that there is ever going to be something like
>> consensus, so I just went and implemented one sensible solution. 
> What if your solution is "not that good" ?, isn't it worth trying to reach a 
> consensus ? don't you think in the end the tool makes the use ? especialy for 
> such type of tags ?
If my solution isn't good it's not going to be used...


> I fear we just have here what I would call the "attribution fear" what would 
> you say about :
> - I create a proposition in your name
> - I use the validate: namespace
> - I correct some.... well... hem most? of it's values to cover your/my needs
> - I do all the blabla and voting stuff
> - I show you the result so you don't need to "spend time on the wiki"
> - we both correct our tools to use it
> - we stop covering other tags usage (yes it sounds like un-democratic, but if 
> we don't, many tags will still be used for the same thing if they are usable 
> in a validation tool)
> - we got our "consensus"
If you can spend the time to find good names for the individual option
(and merge say residential-without-name and poi-without-name to noname)
and do all the stuff in the wiki, I'm certainly not opposed to change
the naming in maplint. If we can reach consensus in this way and have
tags which are not only valid for maplint but for your renderer too,
that be really good.


>> If another tag emerges as consensus it's easy to replace noname=yes with
>> that. 
> Even if we have thousand noname=yes, thousand of internal=noname, thousand of 
> validation=noname in the DB ? not so sure...
If it is that scatter? No. But if we had a couple hundreds noname=yes
and tenthousands of $whatever?



>> With that attidute we'd still have blank map and would be discussing if
>> it should be "highway", "way" or "street".
> Come on ;-) I didn't say that attitude is need for the whole programming 
> world, I say we have a chance, right now, in 1 day, to reach a trade-off, not 
> so bad consensus by spotting problems and make it work.
Sure. If we can find a quick an down-to-earth consensus within the next
couple of days, I'm all for the necessary changes. I just don't want to
get caught in months of discussion, voting and board meetings.


>> As a dinosaur I find mailing list much 
>> more comfortable than wikis. 
> no problems ! I'll copy my end page on the validate: namespace in the talk 
> list if you allow me to. 
Sure. Go ahead.


>>> why restrict that to residential ?, what about :
>>> validate:noname ?
>> Because it is a single test in maplint. For the case of unnamed POIs
>> there is validate:poi-without-name=ignore. (Which is also shut up by
>> noname=yes)
> What do you think of 
> * validate:noname=yes and * validate:no_sign_name=yes
> OR
> * validate:name=noname and * validate:name=no_sign_name 
> (looks familiar eh ?)
> 
> I prefer the second because it makes a street having no name and having no 
> sign with a name impossible. 
> ( If find them exclusives, since if there is no name, there can't be a name 
> sign, or else there is an error) 
> But if some people thing it should be possible, then I'll propose the first
I'd go for validate:noname as a key and "yes, ignore, nosign" as valid
values.



Patrick "Petschge" Kilian