[josm-dev] validator question, multipolygons

[Ulf Lamping]

Am 05.03.2011 21:27, schrieb Dirk Stöcker:
> On Sat, 5 Mar 2011, hbogner wrote:
>
>> We lost some new OSM mappers because of this.
>
> If the people are discouraged that easily then they would have gone soon
> anyway. Have you ever got a message/email from someone who thinks that
> you "destroyed" his work due to a simple modification. The validator is
> harmless compared to this. My father who does a lot more than I do gets
> these messages constantly and I got them a lot when I was more active.

Strange, I never got a single message like this since 2007 ...

> The time for basic mapping is over (at least in Germany and central
> europe)

That's simply untrue, except for big cities.

> and tools like the validator are more and more important to get
> a useable database. JOSM's goal is not to have to ultimate freedom for a
> mapper.

Is this JOSMs goal or your personal?

 > JOSM allows you to do nearly everything, but it does not
> encourage you to do so. The goal is a usable database following some
> unique standards. And it must be easier for a user to follow the
> standards than not to do it.

Again, is this JOSMs goal or your personal?

E.g. it is not the goal of OSM (and it should not be the goal of JOSM) 
to warn about "unknown relation types".

> Maybe in some cases the tests aren't perfectly correct,

The last time I've tested it, the results were unjustified, 
overreacting, not helpful on how to fix it ...

> but this is to
> be expected as automatic detection of these topics is a very complicated
> issue and errors can't be prevented.

As the experienced developer can't handle it, the unexperienced mapper 
has to deal with it - ouch! :-(

> When you find such problems, report
> them properly and they will be fixed when possible.
>
> If I judge this issue based on the ticket reports we get, than we have
> only minor problems with this. And half of the reports ask to add
> additional checks and not to remove some.

If most people just switch off the validator because it's just bogus and 
brocken - which it is IMHO, it's no wonder that there are not much 
tickets about it. Why writing a ticket when you can easily get rid of 
"that annoying stuff".


If you want to have such a validator, it has to be as perfect as 
possible under all circumstances and the messages have to be very clear 
how to fix the issue. Otherwise it's just a waste of everyones time.

Regards, ULFL