"Protection Civile Courbevoie / Neuilly-sur-Seine / La Garenne-Colombes" ne me semble pas être un hôpital - c'est juste leur base... osm.wiki/France/Protection_Civile détaille comment l'étiquetter.

Thanks - the sign wins... I'll alt_name the other spelling from my source. Ask locals if you get there - it'll make a fun chat !

Hello, I originally named Rhazabe (I should have put the accent in - Rhazabé) sourced from a local, and I seen you renamed it Hazabe... Was my naming spelled wrong ? I'm looking for sources to clarify, but all I find is sites using Openstreetmap data - so that doesn't help. You have a better source ? Of course the "H" in Arabic pronunciation sounds like a "Rh"... Does an official spelling exist or are we all transcribing phonetically ?

Sorry for large changeset bounding box: I had started with a simple attribute edit of incorrectly capitalized health infrastructure names and attributes, and got carried away in Kaffrine...

Sorry for the uninformative changeset comment: RAM was saturated and I couldn't edit the field, only select a value. This is of course about capitalization: BWAISE -> Bwaise.

Bien vu - merci ! Corrigé par changeset/165006620

I'm glad you restored relation/12571016 - thanks ! Bert Araali was adamant it isn't appropriate, he reverted it and I didn't want to confront him. How did you decide to restore it ?

Certain du nom "Étoile d'Endynion" ? Cette orthographe est tellement rare qu'elle en est suspecte - et le nom dans la version précédente est "Étoile d'Endymion"... Une source pour appuyer dans un sens ou dans l'autre ?

Take a look at the Osmose "tree intersects with building" - there are quite a few in Kampala. Also, so many false trees that I usually filter them out and don't even bother.

If you can add instances of buildings that you think I may have erroneously deleted, please pass their identifiers - I would like to take a look to improve my mapping.

Lifecycle tagging would make sense for buildings that still exist, even in trace form. I agree that it would be applicable to some ruins here that while no longer buildings are not entirely removed from all physical existence either - I'll consider that.

As for correcting erroneous buildings, it depends: if a mere squaring or a small shift is sufficient to map it, then I do it. But when it has no geometric resemblance to an actual feature (typically sprinkled buildy dust, zig-zag polygons in the general area of a building and the famous #GoMicron copy-pasta garden-hose cartography), then deletion is the most economic way to improve the dataset: zero information is better than disinformation.

Is this definition of wetland foremost an administrative one ? How should wetlands that are not protected_area be tagged ? Or are all wetlands in Uganda considered protected_area by Ugandan legislation ?

In the Rwenzori's, sure - and type=quonset for good measure... But here I'm guessing this could be amenity=shelter

Use of Bing explains a lot of the false buildings - https://imgur.com/a/xuh7cLw shows that all others have congruent current imagery but Bing shows buildings long since demolished and doesn't show newly constructed ones. Buildings traced on Bing imagery and since then demolished as attested by all other imagery sources are faire game for immediate deletion. Also, Bing imagery should be banned from current use - seeing it recommended by tasking manager coordinators is especially jarring (and they don't answer messages either).

Regarding imagery alignment, it seems that, in Kampala (situation of course varies by region - my experience in other countries is different), all current imagery is correctly aligned with GPS traces - see this quick comparison: https://imgur.com/a/xuh7cLw. Mapbox and Maxar are the same capture but Mapbox is strangely blurry, ESRI maybe half a meter offset but I think it is actually a sharpening artefact, Bing is aligned with the three others but laughingly out of date.

Sorry for the tea factory mistake. Can you link to it so I'll have a look ? I have noticed that outside of Kampala (can't remember if along the Fort Portal Road or the Nansana-Busunju Road) the ESRI imagery at high zoom level is older than within the city - so I may have erred that same way.

As for consulting previous editors, it is usually a dead end as they are drive-by editors corralled into "mapping parties" by unscrupulous "community" organizers and their hashtag-heavy contributions are abandonware; I give it a try if they have contributions are recent. You are one of the extremely rare persons from whom any answer is forthcoming - let alone a constructive discussion like we have !

Geodetic markers would be wonderful to clear imagery confusion. Bing imagery indeed has been a terrible source of offset features. At the moment, both Mapbox and ESRI imageries match GPS traces well, as far as I can tell from the Openstreetmap-recorded GPS traces available as a layer in JOSM.

I agree on the difficulty of maintaining relations with large number of members. Even when relation is necessary, breaking it into more relation with fewer members and better locality is more manageable.

Regarding the relationship between fine-grained landuse=* and extensive ones, in Openstreetmap - or at least in major renderers, the smaller ones override the larger ones - so a landuse=quarry inside a large landuse=residential is a landuse=quarry. That way, local efforts at fine-grained landuse mapping are not in conflict with larger ones.

While I salute the ambition of fine-grained landuse mapping, the stunning quantity of unmaintained and false data I have come across in Kampala hints that a gap may exist between such goal and the appetite of contributors (wherever they may reside), which is why I advocate the traditional way of Openstreetmap. But of course, don't let me stand in the way of lofty goals !

"Calibration point" - you mean a surveyed geodetic control datum marker ?

Your identical size and identical orientation buildings copy-pasted with complete disregard for reality are horrible.