I don’t have anything separate from this topic to say anyway. It’s just one of the main reasons I map sidewalks to begin with
Edit: OK but I will elaborate. It would be easy to start using broken: more and there’s not much to say about that other than do it. Pedestrian signals though, that’s an interesting topic for a thread
I’m someone who has been mapping casually in OSM for a while, but have been working more regularly in the last year or two, requiring me to dig deeper into the wiki, ask more questions in the OSM-US Slack, and also examine existing elements.
From that perspective, sidewalks mapped as separate ways have been hugely helpful. I’d previously tried to add and update cycling infrastructure in an area I lived in and found it somewhat challenging to navigate all the information in the wiki required to make sure that I tagged the main road ways correctly for cyclists, while separate paths for cycling were easier to navigate and sort out tagging for. As someone who had been casually mapping for a few years, the only reason I knew OSM even had sidewalk data was that StreetComplete sometimes asked me about sidewalks - the data was nearly invisible to me as a newish mapper.
Then I started doing work in an area where someone had mapped separate sidewalks in some of the neighborhoods, showing me map details I hadn’t noticed before and adding to my understanding of the data model. Separate ways for separate uses, at least in the case of something that is physically a separate piece of infrastructure, has been very helpful for me as a mapper, because 1) I can see that it exists at all, 2) see it on the map where someone else added it and go look at how they tagged items I’ve seen on the ground, and 3) editors like iD can provide guidance and assistance when I can explicitly set my intent to map a sidewalk, as opposed to mapping on the road where it’d be overload to ask me tons of details about a sidewalk.
With that all in mind, I think that not only can we get more detailed information for people who are using the infrastructure itself, but can more easily welcome and train new mappers to add the information at all when the features are explicit rather than implicitly tagged on a road way. I don’t think this is the most important argument in favor of separating ways for sidewalks - I think those arguments have already been made by others - but it’s one more reason for it, in my opinion.
it exists in Poland, which - as far as I know - is in Europe 
And to answer question asked: there is NO global consensus but there is typically a local one and it is better to follow it.
Though local mappers may decide to switch which style is preferred.
Which is preferred often depends on which model is less problematic given local culture, laws and mapping activity.
I do think there’s no reason to move a separate sidewalk once it’s there - I tend to see it just as adding more rather than changing the style
Apart from any conclusion, I am here just to spread the word of mapping sidewalks as separated geometries:
for me there is no polemic, as separate ways is always the best as it allows for richness of information and higher level of detail.
(but still needs to be discussed with local community, of course)
While it is correct that it allows for higher level of details, I’d point out that this in itself does not imply that “it is always the best”. Separate mapping has disadvantages too; e.g. to name a few (without intent to spark discussion about which is better, but just to show that other side exists too):
natural=tree nodes each with its own species=*, height=* and diameter=* is way more precise then one (or a few) areas with just landuse=forest and leaf_type=*”, but except for some outliers, majority of users would probably agree that it is bad idea to that precise in that case). While in the past it was often suggested to disregard this aspect as more detail is often deemed more important, it does cause problems (slower operations, requiring better mobile devices and better servers, discouraging local custom instances due to time and resource requirements to replicate DB etc). Separate sidewalks (and by extension, separate cycleways too) ways are only somewhat less extreme.sidewalk:left=separate, footway=sidewalk, is_sidepath=*, relations etc.) It also breaks 3NF (IIRC database purists out there).Absolutely this.
I don’t follow. For space- or bandwidth-limited environments, the download should already be filtering out data that’s unlikely to be used on the local device. In general, it shouldn’t be more complex to filter out footway=sidewalk in the same step that filters out, say, power=minor_line for use cases that don’t require it.
What is more difficult to filter out is sidewalk tags on the roadway that result in splitting the roadway many more times, but at least the space- and bandwidth-limited mobile software I’ve worked on has always been so abstracted away from OSM’s data model, for performance reasons, that the sidewalk tags and extra splitting have never posed a significant problem.
I don’t think anyone is saying that a sidewalk has to be mapped as a way initially, just as no one is contending that a river can’t be mapped only as a waterway=river to start out with. The question is whether it’s OK for someone else to come back in and increase the level of detail. There are local considerations about how roads are built in one locality or another, but I don’t think preserving someone’s first-pass StreetComplete edits for as long as possible should be one of them.
We can split to new topic if you’re quite interested (but as I said, I just wanted to give few counterexamples, not to get deep into discussion), but the gist is it is for example non trivial (to avoid saying impossible) to follow minutely diffs if you somehow try to automagically conflate highway=footway+sidewalk=footway to nearby highway=residential to add sidewalk* tags to it (sure, you can have two databases and conflate data from one to another using some algorithms, but that is neither resource-free, nor fool-proof, nor trivial to do).
Oh, I did not mean that someone should not modify StreetComplete sidewalk:*=* tags to fully expanded highway=footway way if they see the value in doing it, and local community consensus is that it is acceptable. In fact, I do it relatively often myself when I see the need (e.g. more detailed micromapping of road/sidewalk deviations, separate smoothness, separate hazards/obstacles/barriers, etc) in my community (where both ways of mapping are acceptable).
The post I replied to mentioned the tools that (as I understood it) would semi-automatically convert highway=residential+sidewalk=both to create 3 separate ways just for the sake of it, which I do not find as ideal solution, which is why I mentioned several cons lest people finding this thread later get the idea it has only pros since nobody replied, incorrectly guessing that thus there were consensus about it, and decide to do automated word-wide bot edit or something (yeah, it might sound somewhat far fetched, but I’ve seen things you people wouldn’t believe).
Not to bash the tool or its author! – if you were already intending to convert some area to separate ways for some good reason, and the tool reduces the amount of manual work – more power to you! The point being - having a good reason. If you do not have good reason (i.e. no data would be enriched, just bloated), or if the local consensus is against doing it, then I would usually prefer more simple sidewalk:*=* tags.
The same discussion can be helt about cycleways which goes alongside roads for motor vehicles.
For example I mapped a separate cycleway from a high detailed image. Others said this was wrong, because there was no physical separation between cycleway and car road (separation was a solid line), demanded the cycleway should be removed and tagged on the car highway instead. Even if the wiki says this the correct way of mapping such cycleways, and personally don’t think is a better solution, compared with separate ways.
in this case it was not a separate cycleway, it was a bicycle lane. (And we do not map each lane on carriageway as a separate road)
Your example isn’t necessarily analogous to the sidewalks being discussed here. Personally, I’m quite fond of mapping sidewalk ways because, in my experience in the U.S., sidewalks are almost always physically separated by a curb, if not also a grassy verge, a small drainage ditch, or some trees which I’d like to map. But very occasionally, the sidewalk rejoins the roadway to be separated only by a painted line. Urban planners call this a pedestrian lane, while laypeople might still call it a “sidewalk”. In these cases, I do model it as a sidewalk tag on the roadway, based on the physical separation principle. This nuance applies to cycleways as well, although mappers sometimes disagree about what to do when there’s a very wide painted buffer or flexible bollards spaced too far apart.
It’s also very common in Ireland to walk across the road when the red pedestrian crossing is shown.
It’s also very common in Ireland to walk across the road when the red pedestrian crossing is shown.
I was more thinking about crossing a road at a point that is not a designated crossing point, but what you describe is also common in many places (although typically not allowed).
also common in Ireland. 
(technically illegal within a defined distance of a pedestrian crossing at least)
I think simple sidewalks that are immediately adjacent to the roadway should be kept as attributes. If there is a separation like a verge inbetween then it isn’t a sidwalk, it’s a footpath and should be mapped separately.
If you were in a car and needed to get out you should be able to immediately step onto the sidewalk.
If there is a barrier in the way such as a fence or a hill, then it definitely shouldn’t be mapped as a sidewalk.
When there is a gap and the footpath is mapped, the road should be given sidewalk=no
sidewalk=separate should be reserved for when the separate way represents an actual sidewalk with no gap. I think this should be generally avoided because it creates too many ways too close together for no real reason.
By that logic, all the sections of sidewalk with a railing or narrow verge between the carriageway and the sidewalk would somehow cease to be sidewalks. Would they also cease to be sidewalks where there is a legal prohibition of stopping next to the sidewalk?
Defining a sidewalk only by its utility to a motorist isn’t why people map separate sidewalks. Pedestrian navigation and accessibility considerations are far more important to me.
I didn’t suggest that it’s only for motorists, but it’s one of the main differences. It’s equally important for cyclists and people getting off busses. But if there is anything in the way then it’s not a sidewalk but a footpath.
If it’s mapped as a sidewalk then it should be immediately available along the entire length of the way. Anything else is a footpath
Even if there were a legal prohibition on stopping for cars, it would still be usable in an emergency.
That is a quite unusual definition of sidewalk.
Agreed, “sidewalk” is not defined by (existence or lack of) barriers between carriageway and footway parts (not even by low barriers such as kerbs). I’ve recently consulted some dozen popular English dictionaries for definitions of “sidewalk”, so one may want to look into middle of that (longish) post for dictionary references: Pedestrian lane on the road - #88 by Matija_Nalis