[OSM-talk] RFC: what are empty nodes and how should we use them?

[Peter Wendorff]

Hi.
I'm new to this list, invited to it because I sayed something to this 
topic in IRC.
I'm not answering to Frederiks Mail in particular, but didn't get the 
whole topic, so I cannot answer to the original question simply.

I thought about the history feature the OSM has and wonder, if the 
deletion of empty nodes normaly should be a problem.

Please correct me, if I'm wrong, but AFAIK the OSM works as follows:
- an object is created, normally - but not necessary with attributes.
- an object can be deleted any time
- if a deleted object is touched again by changing it's attributes, it 
will be "recreated" so it's present in the database again.

At least that's something somebody explained me a few month's ago.

If that's the case, I observe:
Deletion of attribute-less objects will never be a problem, as long as 
nobody tries to get information from that objects (not included from my 
point of view).

Of course the tactics like "I make three dots to mark for myself, that 
it needs further work there" will fail in that cases. But what's the 
problem in adding a fixme-attribute to that data?

I think, at long sight we need a kind of atomicity for changes, but even 
today I don't see a problem at deleting empty, non-tagged nodes as they 
contain no useful situation - perhaps except for the one, who created it.

I'm not sure wether empty nodes should be deleted automatically or 
semi-automatically, but there are a lot of bots fixing bugs at tagging 
automatically - for me empty nodes are most useless of all parts.

Please consider: Everything in this Mail is based on the assumption 
about deleting/adding above. If that assumption is not true, forget the 
rest.

Regards
Peter Wendorff


On 16.08.2010 19:29, Frederik Ramm wrote:
> Martin,
>
> MP wrote:
>> You will make some specific check for three continuous dots? Well,
>> there are areas with thousands of such orphaned nodes and trying to
>> check if there are somewhere three dots in a line in such areas
>> wouldn't be easy - either the check will be very slow, prone to
>> errors, or you will need some complex and sophisticated algorithm to
>> make it at least somewhat reliable.
>
> Yes, I think a re-think is in order regarding the validators or at 
> *least* those with a direct influence on editing like the JOSM 
> validator. When they were introduced, people were relatively sure 
> about what they were doing and the validator was just an "ummm, not 
> sure if this is right...?" voice, to be taken with a grain of salt. 
> But nowadays, too many take the validators for gospel, and refrain 
> from making legitimate edits because a validator flags them up. Today, 
> validators should be more cautious about what they flag.
>
> Bye
> Frederik
>