I have to second Martin Constantino–Bodin. Large bounding boxes end up spamming the changeset list when people are examining their local area, even when your changeset has no changes relevant to them.

Also, what makes for osm.wiki/Good_changeset_comments 🙂.
What exactly did you do to these so-called random little islands?

What makes for osm.wiki/Good_changeset_comments 🙂.

Also, smaller changeset areas, please.

Smaller changeset areas, please.

Also, what makes for osm.wiki/Good_changeset_comments 🙂.

Doubtful. This user isn't the only one ignoring the warning dialogue which StreetComplete gives when the user is more than a few dozen metres from the element being changed. One case involved a change in Germany, then a change in Spain, within 30 minutes (he had it set to auto-upload (the changesets remained open), so they were real-time rather than delayed batch). Even a charter flight wouldn't be that quick.

What makes for osm.wiki/Good_changeset_comments 🙂.

Also, smaller changeset areas, please.

Good spot 👍.

“[…] opening_hours:covid19 takes as value the same (machine-readable) syntax as opening_hours […]”

My apologies, the syntax is valid & correct. I had been misled by another wiki page claiming that *:covid19 were the same as the originals except for being a conditional-modifier. Yet, opening_hours:covid19=* makes clear that special comparison values are accepted.

I'm very local. Lots of GNSS tracks are on my to-do (‘wish’?) list along with other types like Mapillary images (once I have the equipment). Some roads have zero. Those that do tend to be major roads (and common driving routes) with only a few separate (in time, or user / device) traces. I, too, have noticed (and been irritated by) the several-metre misalignment of various imagery (with no (nearby) offsets stored in the public DB); seemingly differing by imagery supplier. It's particularly troublesome for small elements.

I did look into applying & uploading offsets based on GNSS (when mapping a small path, for which the alignment was clearly several-metres off in a systemic way), but the GNSS receiver in my nano-computer doesn't seem accurate enough to do so reliably (at best, regardless of software-reported accuracy, it seems to be ~5m). I'm hoping to visit the handful of known survey-points, and find any others (e.g. asking the Planning & Infrastructure departments), as a starting point to get a quantitative handle on the uncertainties involved.
A better external GNSS receiver might be sensible, too.

Since I'm a relative newbie, what would you recommend as strategy?
Major roads by car first, then minor ones, then footpaths?
Broad-strokes across the island, then detail in dense areas, or complete one area before starting another?
I don't really do arm-chair mapping, so I'm unsure what would be most helpful for those who do, or for imagery-alignment.

So far, I've only captured some small tracks of missing features (e.g. the odd footpath while surveying). Though, your mentioning of tracks has reminded me to name+tag+describe+upload my remaining ones. I've not yet done any large captures, though it seems (technically) straightforward & uncomplicated.

Unfortunately, there's only a handful of locals contributing (and it's even worse for the other islands). Not even enough to use the term ‘community’. Yet, self-evidently, a lot of work still to be done.
Thus, I'm thankful for changesets like this, which help a lot.

Ideally, some better (resolution & recent) (& libre) imagery would be good. Small islands are often neglected (memories of fuzzy & very outdated imagery from Google, when they were the only notable source). The catch being that very few locals have even heard of OSM (there isn't really any tech-culture here), and I've only seen a couple of drone pilots (but never managed to speak with them about the possibility) who have UAVs capable of decent altitude while carrying a camera. Chartering an aerial survey by traditional airplane (e.g. a private-pilot Cessna) would be expensive (landing fees, here, are considerable). A balloon would be tricky, due to never being far from the airport.
The States did, some years ago, commission a survey by military helicopter, for purposes of checking compliance with Planning rules, but haven't released that data (and even when they do release cartographic data it tends to be with a proprietary (OS-style, or simply ‘all rights reserved’) license, even for matters of public-interest); I've been meaning to talk to them about that, too. However, someone with your standing might be more persuasive (unless experience shows that it's better coming from a local)?

Any other suggestions or advice would be appreciated. Feel free to PM me (since this is already off-topic for a changeset discussion). So far, most of the input I've received has been from those who are far away (notably Germany), and when these islands have enough non-trivial differences from the mainland, the culture-barrier becomes a factor, particularly when it comes to how to map something unusual.

‘What's best for me’ doesn't seem overly compatible with a collaborative community project.

The obvious counter-argument is that while your time might be saved, this is at the cost of other contributors (many who didn't comment still had to mentally filter-out this changeset).

Quality takes care.

Find a way to automate your desired goal; have the machines do the sub-division work for you. Best of both worlds, then.

building=q (way #963589609 & way #963589668)? I assume a hasty snafu when trying to square the outline and setting building=yes 😋.

Otherwise; more food for surveyors, thanks 🙂. Looks like several-hours worth of work.

The value is also incorrect syntax (according to the wiki); should be “6'6"”.

Besides the key name, the value is also incorrect syntax.

“1 hour” not ‘1 hr’.

Thanks, seems good.

That's the trouble with having so many things (especially roads) named in French, in an island which primarily speaks English; we don't know how to spell the original names 🙂.

PM'd Friendly_Ghost with a pointer to here, since he's busy editing and likely hasn't seen this discussion, yet.

Previous commenters have a good point (and thanks for the hint about timezone=*). However, I'm here to plead for smaller changeset areas.

Rather an enormous changeset area, but fair enough.

Nation ID != language ID 🙂.

See, what did I tell you, Cas?(!) 😄

Remedied (finally) in changeset #108443542.

Re way #206640339

website=* should be a URL, not (merely) a hostname.

Even simply just “add trees” would have been meaningful.