Re: Preset support of alternative preset items

2020-08-19 Thread Martin Koppenhoefer


sent from a phone

> On 19. Aug 2020, at 14:20, Stephan Knauss  wrote:
> 
> alternative: to link to a similar preset with different tags


in general an interesting idea. Maybe the alternative relation could have 
subclasses for more generic, more specific and similar level of specificity 
alternatives
(particularly when the same key is used, i.e. tags are mutually exclusive, 
preference should be given to the more specific tags)

Cheers Martin 


Re: Preset support of alternative preset items

2020-08-19 Thread Stephan Knauss

On 17.08.2020 23:00, Simon Poole wrote:

> One of the frequent use of them in JOSM is for address and contact 
fields.

>
> This is semantically something totally different than suggesting an
> alternative preset item (because the current one might be a common
> mistagging, the alternative might  be a more specific object, etc). And
> because it is semantically different it should be presented differently.

I think I have an idea what you mean. So it would be an attribute 
describing the relation between two linked presets.


How about something like this:



with the relation being one of the attributes:

alternative: to link to a similar preset with different tags
companion: to link to a preset with tags often used in combination
modern: to link to a preset which is a more modern style of tagging for 
the same thing (eg public transport)
legacy: to link to a preset which is a legacy style of tagging for the 
same thing (eg public transport)



The "companion" is already linked frequently in the wiki in the 
Template:KeyDescription as "combination". The "alternative" would be 
covered by "seeAlso".


So we could have this reflected in the presets as well.

Stephan



Re: Preset support of alternative preset items

2020-08-17 Thread Simon Poole

Am 17.08.2020 um 22:40 schrieb Stephan Knauss:
>
> Can you elaborate a bit more in what way you wish this to be more
> structured?
>
> It is there to list similar tags. Do you want to structure the level
> of "similarity"?

To avoid any misunderstandings this is not about individual tags, but
about the real world objects complete preset items describe.

The typical use of linked presets (and where they make most sense) is to
add "sets" of tags that would overload the preset in question  if added
as individual tags.

One of the frequent use of them in JOSM is for address and contact fields.

This is semantically something totally different than suggesting an
alternative preset item (because the current one might be a common
mistagging, the alternative might  be a more specific object, etc). And
because it is semantically different it should be presented differently.

>
>> - add an attribute to the preset_link element, example >     
>> 
>
> So what else besides "alternative" do you suggest?
>
>> - introduce a new element, example
>>      
> I do not see the difference here.
> besides "link" for generic links, this "preset_link" already is a link
> specific for alternative presets. What would a new one differentiate?

See above, that is not actually how the majority of the linked presets
are used. The concept is a recent introduction (if it wasn't, please
explain why the "Similar but different tags:" text is necessary).

>
>
>> - introduce a new element as a container, example
> All references to "similar" presets are already known by the presence
> of "preset_link". As you would have either one "alternatives" in case
> there are links, or zero if not, what is the benefit of another
> container level?
>
Again, see above.
> Stephan
>
>


signature.asc
Description: OpenPGP digital signature


Re: Preset support of alternative preset items

2020-08-17 Thread Stephan Knauss

On 17.08.2020 16:29, Simon Poole wrote:


Some of the recent patches to JOSMs default preset have included
 elements for what are essentially presets for "similar"
objects.

This would seem to be to me quite interesting information to capture,
however the specific unstructured way of handling this would seem to be
a bit sub-optimal. 


Can you elaborate a bit more in what way you wish this to be more 
structured?


It is there to list similar tags. Do you want to structure the level of 
"similarity"?



- add an attribute to the preset_link element, example >     


So what else besides "alternative" do you suggest?


- introduce a new element, example
         

I do not see the difference here.
besides "link" for generic links, this "preset_link" already is a link 
specific for alternative presets. What would a new one differentiate?




- introduce a new element as a container, example
All references to "similar" presets are already known by the presence of 
"preset_link". As you would have either one "alternatives" in case there 
are links, or zero if not, what is the benefit of another container level?


Stephan