Or maybe shop=bed bed=mattress makes more sense ans such cascading tagging to indicate that shop is only/primarily about mattresses would work well?
Any other ideas? Or maybe I am trying to be overly detailed and I should just use shop=bed without attempt to tag actual shop type?
Maybe mattress-only shops are local aberration not known elsewhere?
My plan and idea for now (unless this thread will convince me otherwise) is to document shop=mattress at wiki as valid and making sense and retag mattress shops nearby to it.
So mattress= was mass-added in 2022. But bases= should be removed until clarification on who used the 21 instances, being mistranslated for frames. linen= is not an accurate word for sheet covers, and bedding, confusing whether it’s a “metonym” for them, or the specific materials sold. Tag:shop=bed - OpenStreetMap Wiki =bed seems it could be misnamed, eg =bedroom products/essential/furniture. It’s too late now.
But besides mattresses, there are dedicated pillow shops. Does that mean the problem is not shop=mattress ? It’s how to describes shops selling mattresses and bedding, without frames and whole beds.
I’d go with shop=bed, (and maybe extra tags like mattress=yes if you like.)
I personally just add description=* giving information about details what is being sold / services being offered when shop name is not precise enough.
Having instead to add multiple nodes (e.g. shop=mattress + shop=bed_frame + shop=pillow + shop=waterbed + shop=linen) at the same location (with duplication of all other data, i.e. name, opening_hours, website etc) in cases when it offers all of those (which is somewhat common situation) would sound like a much worse solution to me (compared to slightly unspecific shop=bed).
I already have to that as least-worst-alternative for shop=stationery + shop=books (which are often combined here), or shop=copyshop + shop=photo (ditto) and I quite heavily dislike it. [1] I certainly don’t want more of those.
I don’t think that anyone is suggesting that are they? The question is how to tag a shop that “only sells mattresses” compared to one that “sells beds (including both bed frames and mattresses)” etc.?
There are a couple of ways of doing that - shop=mattress for the first case and shop=bed for the second, or shop=bed with subtags for both. You can argue either way, but taginfo shows an immediate problem - mattress is hidden among typos for beds=!
What shouldn’t happen is the loss of data described above.
I am interested in how to tag shop selling specifically/primarily mattresses and not one selling also bed frames and pillows and linen.
Instead of shop=mattress + shop=bed_frame + shop=pillow + shop=waterbed + shop=linen I would use currently available shop=bed (maybe with mattress=yes + bed_frame=yes + pillow=yeswaterbed=yeslinen=yes or similar tagging)
I agree with that (shop=bed being overencompassing tag). So similar to that how you would tag that case with multiple things being sold, I would also suggest similar tagging if the shop happens not be selling bad frames (e.g. shop=bed + mattress=yes + bed_frame=no + pillow=no + linen=no – instead of shop=mattress)
Otherwise (i.e. if we add shop=mattress which overlaps with shop=bed) we create extra confusion IMHO – what if that mattress-selling shop (but no frames) also sells pillows (somewhat common combination over here)?
if shop=mattress is defined exclusively as a shop selling only mattresses and absolutely nothing else, then one would have to use shop=bed + mattress=yes + pillows=yes instead; which makes the new tagging problematic, as now data consumers wanting to buy a mattress have to look for multiple primary tags (one for shop=bed, and other for shop=mattress) instead of just one that they had before (shop=bed) – IOW, we’ve just made everyone lives more complex for no real benefit.
if shop=mattress is however defined as a shop only primarily selling mattresses, but also possibly other related things, then one might tag it as shop=mattress + pillows=yes. But they might also sell linen, so add + linen=yes. And waterbed is a kind of mattress anyway, so waterbed=yes too can be added to it. And then, many which are not selling bed frames but only mattresses also sell (similarly rolled up) bed bases. So add extra bed_base=* if they also sell them? See the direction were this is going? Now we’ve practically made a duplicate of shop=bed but only with implied mattress=yes.
But even for such such expanded shop, since they’re not selling bed frames[1] mapper could be reluctant to tag them as shop=bed (for the same reason you’re reluctant to use if for mattress-only shops)
So in that case, it would bring more confusion to mappers (and there would be plenty of one tagged as the other, and vice versa; again forcing the data consumer to do multiple searches until the situation was cleaned up, which would likely never going to happen).
And it escalates. E.g. if shop=mattress makes sense for matress-only (or mostly) shops, then there is a case for shop=pillows (also somewhat common here). And then how the shop which sells both mattresses and pillows (but no beds i.e. bed frames) would be tagged?
That I agree with!
TL;DR: I think shop=bed + mattress=yes + bed_frame=no + pillows=no is much more usable than shop=mattress; both for mappers, and especially for data consumers.
That would be a poor definition (similarly, shop selling variety of fruits and vegetables and also making possible to buy public transport tickets can be tagged as shop=greengrocer)
“only/primarily about mattresses” was in the initial post
obviously, it it sells all these things it fails “only/primarily about mattresses” and should be tagged as shop=bed or other wider tag
depends on what is dominant product sold there (and yes, for borderline cases two taggings will be possible - this is normal and present already)
As an aside, here is a list of the shop tags that the raster and vector map.atownsend.org.uk maps can deal with.
The most popular ones get their own icon (about 60 different ones for shops, but most are “variations on a theme” to avoid confusing people too much).
There’s an amalgamation of “similar” tags into one value - 7 different things get the “convenience” one, and 45 get “furniture”. 119 of the 450 get a “shopnonspecific” fallback dot.
The amalgamation chosen here is one that mostly works for a general purpose map, based on shop values present in UK/IE (which doesn’t include mattress or similar). A map for a different audience might want to do something different - perhaps many more shop icons, perhaps many fewer, perhaps the same number but different categories. As long as the data’s present in OSM (either as different shop tags or different subtags) it doesn’t matter to me. What would matter is where the OSM tagging is changed so that I can’t choose which category to use because information is lost - but neither of the suggestions at the top of the thread would be a problem.
In Shrewsbury there is a shop called Dial a Mattress although it does also sell beds although mattresses are replaced more often than beds so a shop specialising in mattresses seems very feasible.
Somewhere selling public transport tickets other than a public transport company is certainly weird in my experience. I did get caught out by this in Trieste when I found I couldn’t pay on the bus.
Yeah, which is exactly my point – that it is confusing as “only” has a different meaning then “primarily”, and it is unclear which one gets more priority, i.e. where the line in the sand is drawn.
If it is 99% mattresses and 1% pillows, it is (hopefully) clear that it would still be shop=mattress. But if it is 95% vs 5% ? 90:10 ? 80:20? 70:30? 60:40[1]? At which percentage does it stop being shop=mattress and should fall back to shop=bed?
IOW, such tagging as shop=mattress would:
introduce extra ambiguity and needless brain power wastage for mappers [2],
while at the same time making life more complicated for data consumers (need to search for two separate tags instead of single one).
While OTOH staying with shop=bed + mattress=yes + bases=no + linen=no + furniture=no is:
already in use
needs no changes in mappers practice (no re-education effort)
it is very clear for mappers what to tag, so no brain power wastage nor ambiguous boundaries:
if it does not sell mattresses at all, add mattress=no to shop=bed;
if it sells only mattresses, add mattress=yes + *=no
if it sells combination of mattresses and waterbeds, add mattress=yes + waterbed=yes
etc.
is simpler for data consumers (single search is enough if you want to buy mattress; or anything else)
does not require extra primary tags, and easily expands without much controversy for other use cases e.g. shops selling primarily pillows.
Nice example. So how would people tag that specific shop? As shop=mattress, or as a shop=bed? I’d wager that opinions would differ.
What might work for both “sides”, is perhaps adding an iD preset (instead of inventing new tagging which overlaps with existing tagging).
I.e. make an iD preset called “Mattress shop”, which tags shop=bed + mattress=yes
(people can then tag it easily, but also they can easily add other tags without making a conflict, e.g. waterbed=yes if it also sells waterbeds in addition to mattresses)
If that is unclear in some contexts I should try to stop using it.
I meant that it is also fine if shop is primarily about mattresses
95%, 90%, 80% are still “primarily” to me, for 70% or 60% it seems borderline and either would be fine.
But I am not planning to define exact % in the same way as we are not defining exact % as border between various other shop type (shop=pastry vs shop=bakery for example, shop=alcohol vs shop=beverages, shop=chocolate vs shop=confectionery and more often have borderline cases)
but it would enable distinguishing mattress shops from ones primarily about other things
For survey based questions? 100%
yes, allowing to tag mattress shops would introduce a bit of complexity, I disagree that this is a waste of time
I am NOT a fan of this, because then shop selling 99% mattresses and 1% for linen and pillows is not a mattress shop anymore (so you cannot really make presets for mattress shop anymore)
Hmmm, I’m unable to reconcile those two statements:
In former, you seem to accept that while for you e.g. 60% mattresses would still be “primarily mattresses”, others might have different evaluation (which I too think is unavoidable).
Yet in the latter, you seem to say (i.e. that 100% quote) that everyone will always make the same determination that this “60% mattresses shop” is always a shop=mattress (at least that is what verifiability means to me)?
Could you clarify which you meant?
It seems I was misunderstood, I did not mean to imply that adding detail to a shop so we know it primarily sells mattresses is a waste of time. In fact, I find adding such detail is quite useful!
My complaint was that newly suggested tag shop=mattress is inferior solution, as it requires users to overthink whether they should should choose shop=bed+mattress=yes or shop=mattress, i.e. “waste time deciding which choice to make” (e.g. more in line of xkcd #309 “very bad” case, but involving only a single person).
While I can imagine solutions that might help there (e.g. use linen=limited[1] to indicate that it does not detract from main values specified with yes), I think bigger issue is that it is unclear to me: What problem exactly are you trying to solve here?
If we’re thinking about potential mattress buyers, why should they care whether some shop is selling (almost) exclusively mattresses, or whether they also have (say) good selection of pillows (in addition to good selection of mattresses)?
Would some of potential buyers decide to prefer one (or avoid the other), and if so, why?
(Or is this proposal not at all about catering to potential mattress buyers, but to some other category? )
Only thing I as a consumer might want to know is how good is their selection of mattresses (e.g. if I want to try out several dozen models before making up my mind what I want), but that would unfortunately not be solved by shop=mattress.
As an example, I bought my current one at a small shop selling only mattresses, but they had exactly 2 models to offer (PU foam alone, or PU+memory). It would’ve definitely be shop=mattress, but selection would be very low. Luckily I know what I wanted so I bought it cheaper there, skipping at least two layers of middleman.
On the other hand, some of the shop=bed + mattress=yes + linen=yes + bases=yes + pillows=yes + furniture=yes + waterbed=yes had much bigger selection of mattresses than that shop=mattress of mine, even if they were not “mattress-exclusive” shops.
Maybe something like”shop=special” and “sold_items=mattresses,pillows”. Otherwise I see the danger of a “combinatoric explosion” (e.g. cigarettes and beer, etc.)
The concept of verifiability in OpenStreetMap essentially means that another mapper should be able to come to the same place and collect the same data (“verify” the data you have entered).
At the core, “verifiability” is that everything you do can be demonstrated to be true or false by other mappers – the latter hopefully implying that there has been a change on the ground that needs mapping.
there are some borderline cases (shop on boundary between shop=bakery and shop=pastry) where data is verifiable, but different mappers may judge it differently. You can find some cases where half of mappers would tend toward A and half toward B.
The same for tracktype, surface and so on. This is annoying but gets far less annoying when you realise that for very borderline cases selecting either is fine.
Isn’t shop=bed already distinguished from other shop types (e.g. shop=convenience)?
But my question was not about why should shop=mattress be findable/shown/rendered differently than any shop (i.e. shop=*) which you seem to be talking about?
(and where the answer might be something like “because I want to go buy a mattress, and not bread and milk, and I prefer to know to which shop I need to go, and for that I need to distinguish shops selling mattresses from shops selling bread and milk”).
But my question was instead Why should shop=mattress be findable/shown/rendered differently then shop=bed+mattress=yes?
What would be the answer to that question?
[click here] for an aside about verifiability, but not that important for main discussion
Yeah. I would read that quoted verifiability page exactly to mean that distinguishing A vs. B is not verifiable (i.e. if it were StreetComplete resurvey quest, we’d get a very long history where each person would tag it with different tag and cycle repeats)
The fact that downside of that unverifiability is more tolerable in some cases than other is of course important consideration factor, but does not IMHO change the fact that it is not verifiable. It just means that the consequence of that unverifiability is quite tolerable.
But OK, I think that is was cleared enough for me (it seems we just interpret the term to mean something somewhat different, but agree on the end result; i.e. that small amount of unverifiablity [or whatever we want to call it when two people tag the same thing differently) are usually fine, while large amounts usually aren’t)
Sure, if we already have competing interpretations, such realization that it is not a big deal makes those aggravations less annoying. But I still would say we should strive to avoid introducing yet another such annoyance if we can help it.
)