May 16, 2023, 19:22 by ajt1...@gmail.com:

> On 24/04/2023 16:57, Mateusz Konieczny      via talk wrote:
>
>>
>>
>> Apr 22, 2023, 14:10 by >> ajt1...@gmail.com>> :
>>
>>
>>> More generally, anyone with half a brain consuming OSM shop          data 
>>> (or actually, _any_ external data from _anywhere_) will          look at 
>>> the values contained in it***.
>>>
>> And that is exactly what lead to proposing this        edits - I was writing 
>> code to handle OSM
>> data and researched tagging situation. And one        of[1] effects was 
>> discovering numerous
>> cases of tags that seem to be exact duplicates of        more standard ones, 
>> and retagging
>> them seems to clearly improve OSM data as far as I        can see
>>
>>
>
> Your continued tagfiddling here is making it much harder for      local 
> mappers to find problem values in OSM data.
>
>
> No-one's going to complain about you changing "shop=shoe" to      
> "shop=shoes" - they clearly have the same meaning, so changing the      less 
> common form to the more common form is a net benefit.
>
>
> However, your recent changes have gone much further than this,      included 
> changing shops with values you don't understand into      "shop=yes".  
>
>
For start: no such changes will be made, if requested I can revert changes
of shop=fixme to shop=yes if it is disputed.

-----------

But I do not really agree with some claims made here and want to explain why.

For start there is a long list of shop values which meaning I do not understand
(for example, from start of list of exactly such values:
shop=grossery, shop=towing, shop=showroom, shop=salon, shop=garage,
shop=pond, shop=consignment...)

This change was made because it was carrying no real info and was obscure value
unlikely to be found and handled by mappers.

> As an example, consider > https://www.openstreetmap.org/way/353944525>  .  It 
> was previously      "shop=retail", an unusual and rare tag that would likely 
> flag up      the interest of a passing local mapper.  You changed it to      
> "shop=yes", of which there are 180,000 of in OSM.  
>
Except people using JOSM validator*, iD, StreetComplete and other tools with 
special
support for shop=yes as feature tag. Note that this tools are more or less 
proactive
while shop with unusual values will be spotted solely by manual checks
(as listing all unusual shop values is far from helpful and requires a long
filtering to skip undocumented values, aliases, confusing values carrying some 
info)

*and therefore also Osmose, though I would not recommend this QA tool

> No-one is going to spot that as an "unusual" shop at all.  
>
Thousands/tens of thousands were fixed already, at least 10 000 would be 
definitely not spotted and not fixed if they would be just one of 10237 rare 
shop values.


> This one's actually a      charity shop, and a question about it on IRC would 
> have got that      response in only a few minutes (or a glance at 
> taginfo/overpass: > 
> https://taginfo.openstreetmap.org/tags/name=Lighthouse%20Charity%20Shop#overview>
>   / > https://overpass-turbo.eu/s/1v3j>  ).
>
> Changing "invalid" (by whatever definition) values to "valid but      
> incorrect" ones does not improve the quality of OSM, and it      actually 
> hides problems so that they are much harder to fix in the      future.
>
>
I would describe this change as "invalid unusual value" to "invalid common 
value,
supported by QA tools and more likely to be fixed or specially supported"

>
> In the case of the changesets that I've seen just now and      commented on 
> (see > https://resultmaps.neis-one.org/osm-discussion-comments?uid=3199858>  
> ) many are by longstanding contributors to OSM.  In many cases a      comment 
> on a previous changeset would have got the answer "yes,      obviously that 
> should be a shop=xyz" (rather than you just setting      it to shop=yes). 
>
>
I am doing also this - and maybe opening notes/making changeset comments first
in all such cases would be preferable.

>  I thought that after the discussion on > 
> https://www.openstreetmap.org/changeset/134837986>  that you weren't      
> going to mass-change actual values to shop=yes any more, but      clearly I 
> was wrong.
>
So you would consider any and all changes to shop=yes as making things worse?


> (for the avoidance of doubt, writing in a personal capacity)
>
the same

_______________________________________________
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Reply via email to