It is unreliable when you want to determine “type of factory/industrial facility” based of it, as a data consumer in large amounts.
Because they’re simply part of the facilities’ common names “paper mill, sawmill, rice mill”, or to clarify the activities occuring at the tagged facilities. plastic_manufacturing facilities and plastic_processing facilities are significantly different from each other.
The highlighted text is also exactly what I tried to do with my changes to industrial page.
This “seperate key for type of factory/non-factory industrial” idea has already been proposed here, the only difference there being that the new key would be called “works” for type of factory (as subtag for man_made=works).
I am not especially opposed to this, but I find the usefulness of this separation (industry for "type of production facility/factory, industrial for non-factory industrial establishment) questionable. Also, it would be a much larger impact to existing tagging than my orginal changes to the industrial page, also because most current industrial uses are on factories.
What I mean is to create a new key (industry) that is basically industrial=* version 2, using common and generic words for types of productions facilities. Then we can easily categorise them, translate them and users can generate labels for them. product can then be the more specific information, like the type of produced food at a food processing plant or the types of oil products made at a refinery.
I believe such ideas should be presented in a proposal rather than just editing the wiki. Just noticed some recent wiki fiddling where documented and used product tags were marked deprecated in favor of industrial=same value here, which neither weren’t discussed nor based on clear usage numbers.
I strongly agree. It would be nice to end up with a good proposal for this. At this point any unilateral changes are counterproductive and whoever made those changes should receive a link to this discussion.
I would tend to (carefully) agree here, but, as I said along with several other posters, I still think that needs a formal proposal, if one of limited scope. The proposal (or a series of proposals) would still need to establish:
A relatively complete overview of the scope of industrial tag. For instance, the whole key:industrial as proposed significantly overlaps with man_made=works – it seems we should eventually deprecate the latter since it’s redundant.
A proposal how to mark storage facilities. They currently lack a consistent top-level tag. For example, we have warehouses; open-air depots (timber; gravel and sand; coal; machinery; ore); agricultural silos; refrigerated warehouses… I envision something like industrial=depot (to broaden the current vehicle-focused definition) and/or industrial=warehouse, accompanied with a resource=*… or product=*… but to be discussed.
A proposal how to mark extraction facilities: we have landuse=quarry for open-pit mines (not the best but…). However, man_made=mineshaft looks too narrow for a typical pit mine containing several shafts and a significant supporting facility (storage + machinery depot + processing). And then, the whole oil and gas field business…
Last but not the least, at least an overview/acknowledgment about “other” industrial facilities. For example, recently I encountered an industrial-scale laundry washing plant (shop=laundry; access=private?! ). Surely there are other, gray-area facilities such as those; water works come to mind.
You find it unreliable, when their usage haven’t been defined clearly until now. Is it a fair comparison with your proposal? You are concluding too soon. To make a disclaimer, using product= for specific items is something discussed here, that may well be one possibility only.
Being a common word doesn’t mean you should always use them in full. =oil_pipeline or =oil_tank is not used when there is substance=oil and content=oil . That applies to product= and industry= here.
On a related note, your proposal still has eg =refinery not =oil_refinery , unlike =sugar_refinery . What about natural gas and metal refineries? Why should =gas_plant and =mineral_processing be used? Where is oil storage when there is =gas_storage . Same as my question about pet food and food manufacturing, the options to be considered for this industrial= is endless. On the other hand, this can be organized with product= , industry= , and something more specific.
Plastic synthesis and processing is again a matter of how to format this info. As a starting point, what does product=plastic mean? If it’s bad, it can be changed to eg =plastic_material and =plastic_product in industry= (instead of industry=plastic if it matters) , or somewhere else. Or as I said, use industrial_process= . Then should a plastic bag factory be =plastic , =bag , or =plastic_bag somehow? What happens when it makes bags of different materials? How should the product function and composition be structured? All these are issues to be considered, that are not solved by calling it done at =plastic_manufacturing and =plastic_processing .
As a note, similar to how there are classification systems for industries, standards exists for goods too. There’s WCO’s Harmonized System in trade that may be referenced for products.
This is what I think too - multiple individual proposals for more specific topics/problems can produce better solutions, than just one giant proposal for “everything”.
However to be clear: I don’t want to be causing any large changes to tag usages (and I’m not doing so with my draft page, as demonstrated), I only need a documented way to tag facilities consistently (or everyone who wishes to map those). I removed all things which could change the way the key is used from my draft page, and I (almost) only used values which were already in use, fitting the definitions I listed.
Yes, the way “product” is currently documented & used - it is unsuitable for tagging the “type of facility”. I also don’t know how this should be documented so that “type of industrial facility” can be determined from it.
An oil pipeline or an oil storage tank is not what is/ever was supposed to be tagged using “type of industrial facility/establishment” industrial. Those are individual industrial objects, which are currently tagged with the man_made key.
Because industrial=refinery is already documented as being an “oil refinery”, it has +500 uses, vs “oil_refinery”, which has 4 uses, when people say “refinery”, they usually mean oil refineries. Also again, I’m not here making up random values, I tried to cover all types of industrial facilities using existing values, wherever possible.
“Natural gas refineries”/“steel refineries” do not exist - you probably mean industrial=gas_plant/industrial=steel_mill, both documented on my list, those are the common names for such facilities.
“gas_storage” refers to underground gas storage (pretty common, over 400 in the US alone), for “oil storage”, there is “industrial=petroleum_terminal”, those facilities are very rarely just “storage”, and they’re rarely just for (crude) oil. There are 4 goverment (+1 private) underground storages in the US for crude oil (don’t know about other places), but that is obivously something very rare, and I wouldnt worry about it for now.
product=plastic dosen’t says anything about the facility, as I’ve said, raw plastic manufacturing and plastic product manufacturing are very different types of facility. What is the point of using something currently entirely unused and undocumented like “industry=plastic_material” over “industrial=plastic_manufacturing”?
It should be man_made=works, industrial=plastic_processing + product=plastic_bag. When it makes bags out of different materials? Very unlikely from my experience, and if that case should ever arrise simply use the primary type/occupation of the facility.
Absolutely nothing wrong with considering those points! - I’m not “calling it done” anywhere. But we gotta start somewhere, with some tagging which makes a reliable tagging of the most basic types of industrial facilities even possible, and those basics shouldn’t need to include the most special case somewhere deep in the mountains.
classification systems for products, sounds interesting, but I’m here to tag industrial facilities, and note that good information about the exact products produced is often much harder to obtain than the type of industrial establishment.
Can be considered of course, but those have all already established ways of tagging, and once again, I personally don’t want the big tagging revolutions, I only want to tag features which can’t be tagged reliabily right now.
(If someone else wants to consider large tagging changes, go ahead, I have no problem, but I’m not going to do it in any way, that’s what i’m saying)
You aren’t answering my question. Why does man_made=pipeline + substance=oil and man_made=storage_tank + content=oil work instead of man_made=oil_pipeline or pipeline=oil_pipeline , while man_made=works + product= / industry= / whatever doesn’t, requiring your list of intertwined industrial= ?
Why do you need industrial=plastic_processing with =works + =plastic_bag ? What info does a “plastic bag factory” not convey? That’s the exact industrial facility already. Categorizing based on plastic processing for the industry or process is added info. This doesn’t require, and doesn’t fit industrial= .
They are just some examples. Why should =steel_mill be used, not =steelworks or =steelmaking both also more numerous? OpenStreetMap Taginfo Yet again, the industrial= terminology is many, while product= or whatever =steel is a simple solution.
You are refusing to look at other alternatives by claiming other solutions as “unused and undocumented”. How many of the industrial= are added by you, and others after the wiki page was installed with your content? I casually looked at a few, and your name is everywhere. Well then, they were also `unused and undocumented" before you started using them. Is this a valid argument?
product=plastic_bag is the product, and not the type of facility. industrial=plastic_processing does not only include “product=plastic_bag”, it also includes product=
Plastic water bottles
Plastic food containers
Plastic straws
Plastic cutlery
Plastic packaging
Plastic cups
Plastic plates
Plastic utensils
Plastic storage bins
Plastic wrap
Plastic film
Plastic trays
Plastic tubs
Plastic jars
Plastic jugs
Plastic shampoo bottles
Plastic toothbrushes
Plastic combs
Plastic hairbrushes
Plastic razors
Plastic soap dishes
Plastic shower caddies
Plastic laundry baskets
Plastic hangers
Plastic clothespins
Plastic dish racks
Plastic cutting boards
Plastic measuring cups
Plastic mixing bowls
Plastic colanders
Plastic strainers
Plastic ice cube trays
Plastic food scoops
Plastic funnel
Plastic kitchen gadgets
Plastic shower curtains
Plastic bath mats
Plastic toilet brushes
Plastic trash cans
Plastic recycling bins
Plastic storage containers
Plastic pet food bowls
Plastic pet toys
Plastic pet waste bags
Plastic garden pots
Plastic lawn chairs
Plastic building blocks
Plastic pool toys
Plastic beach toys
Plastic sports equipment
Plastic frisbees
Plastic balls
Plastic bats
Plastic rackets
Plastic helmets
Plastic protective gear
Plastic water guns
Plastic puzzles
Plastic board games
Plastic playing cards
Plastic decorations
Plastic balloons
Plastic banners
Plastic tablecloths
Plastic disposable cutlery
Plastic disposable plates
Plastic disposable cups
Plastic disposable tableware
Plastic party hats
Plastic gift bags
Plastic wrapping paper
Plastic tape dispensers
Plastic desk organizers
Plastic file folders
Plastic binders
Plastic report covers
Plastic document sleeves
Plastic rulers
Plastic scissors
Plastic staplers
Plastic tape dispensers
Plastic pencil sharpeners
Plastic markers
Plastic highlighters
and beware if someone doesn’t add the “plastics” in front of it, you couldn’t even guess the type of industrial facility! (Not that “guessing” wouldn’t even remotely be a good idea)
If a data consumer wants to look for plastic processing factories, should he really be expected to look for all those values, and thousands more, instead of just one value?
Or industrial=plastic_manufacturing does not only include product=plastics or product=polymers, it also includes product=
Polyethylene
Polypropylene
Polstyrene
Polyvinyl chloride
Polymethyl methacrylate
Poly(ethyl methacrylate)
Polyacrylic acid
Polyamide
Polybutylene
Polybutylene terephthalate
Polycarbonate
Polyetheretherketone
Polyester
Polyimide
Polylactic acid
Polyoxymethylene
Polyphenyl ether
Poly(p-phenylene oxide)
Polysulfone
Polytetrafluoroethylene
Polyurethane
Polyvinylidene chloride
Styrene maleic anhydride
Styrene-acrylonitrile
Tritan copolyester
Acrylonitrile butadiene styrene
Ethylene vinyl acetate
If I would start listing possible product values for industrial=chemical, this forum would probably run out of disk space.
Ten posts ago, the complaint was that only industrial facilities with “special names” should be tagged. Now I used the “special name”, and it is another complaint…
Both those values are uncommon names, so I used the most common name.
This “alternative” was “As a starting point, what does product=plastic mean? If it’s bad, it can be changed to eg =plastic_material and =plastic_product in industry= (instead of industry=plastic if it matters) , or somewhere else.”
I don’t see how this is better than industrial (it appears to be less clear also), and yes, “industry” has 400 uses and isn’t documented anywhere. What’s wrong with this reasoning.
I added many to OSM, obivously, since I’m here to map. Still, only a small part of values weren’t used before being added to this page.
As a layperson, I would be surprised to see this broad interpretation of “plastic processing factory” positioned as a top-level type or kind of factory. I’m sure there’s a good reason for doing so in some contexts, such as environmental analysis, but to me it sort of seems like relying on amenity=restaurantrestaurant=tomato to identify Italian restaurants by their ingredients.
At the same time, I appreciate the distinction you’re drawing between the processing of raw materials and manufacturing into finished goods. (Maybe assembly too?) If we really need a key to indicate the materials that a factory consumes to produce the product, then can it be something more direct than industrial=*, like factory:input=*?
As I said, if you want to categorize them, something else eg industry= and industrial_process= can be used. Adapting industrial= for a mix of different levels and perspectives is not less of a mess than the status quo. The usage of product= can be decided, and the missing companion for either the generic or specific item can be created.
How is “steel mill” special? If it’s =foundry or similar for some facilities, I could agree. But my suggestion is to put them in works= , not industrial= . Again there are more worthy distinction to make than simply =steel_mill, viz integrated mills, and mini mills. My criticism is these industrial= offer no advantage over =works + whatever =steel , and there are too much variation in terminology. How did you decide “steel works” is less common anyway?
Are you sure it’s "a small part of values weren’t used "? That’s already false for the prominent =steel_mill . There’s double and a magnitude more =steelworks and =steelmaking . =steel_mill only appeared in late 2021 from you Way History: Cartersville Steel Mill (663982770) | OpenStreetMap , and the rest are added in 2023. I can totally investigate every one of them in your proposal.
I have indeed thought about using factory:output= (following power= ) somehow with product= and other ideas. Unfortunately this does not help solve the proposal author’s insistence on the presumed issue of how factory:input= , product= , and others may be arbitrarily decided as either some generic or specific items, while this very problem has been discussed repeatedly in attempts of making a definition.
To clarify, you have a typo reversing them. The author defines =*_manufacturing as forming the compound (eg plastic pellets) from other raw materials. =*_processing is turning this “raw plastic” / “primary form” of plastic into the finished plastic product.
I’m exactly calling for them to be clearly separated out from the mixture in industrial= , hence my suggestion of industrial_process= among other things. Rigorously speaking, I find “manufacturing” not distinctive enough. When I search “plastic manufacturing” . at least half the results are referring to the entire process from not-plastic to plastic products, and some are talking about =plastic_processing . That’s why I raised an example =plastic_material by referring to the classification standards as a small improvement, but it’s not what I prefer overall.
An issue here is that some companies do several of these steps in the same place, which in turn complicates tagging for us.
I’m not sure if any mapper is going to ask a factory operator or manager for a list of their in- and outputs. Generic tags are probably going to be more successful.
I always found man_made=works completely redundant. Works sounds like a large mechanical contraption. There is no need to specify that it was man made.
I would use it to represent area of industrial work like a sawmill. The product key seems reasonable for the works output.
We do need something to represent individual machines on an assembly lune or on just sitting on the factory floor. Maybe “machine” as it is a purpose built device. So a board plainer as machine=plainer. The larger factory should have a industry=lumber to indicate the type of products produced.
To be clear, I agree that the tag isn’t ideal (neither the key nor the value), but that is what’s currently established, documented, and has significant use, and changing it should therefore not be done lightly.
It’s a discussion about the proposed industrial= . The author was not entirely documenting the existing usage. The edit even changes it from a landuse=industrial attribute to a standalone feature. What you didn’t check is how both =plastic_processing and =steel_mill are created and used by the same user recently, the latter further ignoring the existing uses of =steelmaking and =steelworks (would be fine if documenting own usage, not pushing a proposal by rewriting unilaterally). Not to mention the dismissal of man_made=works and product= as possible solutions, that are used now. They are alternative solutions to the proposal, which I see as mixing everything arbitrarily in industrial= .
On a related note, while there is not much other pattern that can be identified in industrial= for plastics, anecdotally there are 3 https://taginfo.openstreetmap.org/tags/product=plastic_products , though understandably no product=plastic_material .
Most of these suggestions are an attempt to negotiate with the proposer. In other words, to the extent that the proposed system has merit, maybe it should complement the industrial key rather than supplanting it. Then if it’s really unlikely that anyone other than an imported dataset knows a factory’s input and output materials, at least that doesn’t stop us from categorizing the factory in broader terms.
I don’t know if it’s really that broad. From their function, the industrial process is always very similar for almost every type of plastic product: Plastic granulate is delivered, liquified by heat and then “put into form”, for example by injection molding or film blowing.
What do you mean by “different levels & perspectives”? I’d say the values on my list are quite consistent with their scope.
If the “type of factury/manufacturing facility/works” should not be stored in the industrial key and a “works” key or so be used - what is the benefit of using this other key? Most current industrial tags are on factories, and this “other key” would deprecate those uses.
"steel_mill" has 6.320.000 hits on google, "steelworks" has 4.730.000 hits and "steelmaking" has 3.860.000 hits. The difference is there.
This is an important point. No one should be required to deeply study the wiki pages or work there before being able to use those tags correctly. Tags like “factory:input” or “industrial_process” sound like good supplementary information in some cases, but nothing that should be required to reliably tag “type of industrial facility”. Even “product” can be surprisingly hard to research sometimes (unless one is being very general of course).
All mentions of this have already been removed from my version.
I am in no way modifying the use of man_made=works or product in my version. If you want to use product for type of industrial facility - then where does the product information go? industrial is a supplementary tag to those now, like it always was.
To change the pace a bit I decided to just start improving the map by retagging factory=* tags to the proper product=* tags, adding other tags like man_made=works where necessary.
Edit: there are too many for me and some cases are a bit more complex, so I made a MR challenge for us to improve the map together while we think of long-term solutions for factory tagging: Retag factory to product tags
). Surely there are other, gray-area facilities such as those; water works come to mind.