and for completeness, the "missing" polygons jumped back into the database this morning - for example the historic Anglican parishes:

1228a1229
> -10742937 | St. Peter's Parish
1368a1370
> -10709157 | Stillorgan Parish
1416a1419
> -8851833 | Stillorgan CofI Parish 1900
1469a1473
> -8811305 | Taney Parish 1900
1603c1607
< (1600 rows)
---
> (1604 rows)

OK, as requested by PM, I've reverted this changeset and the related one before it.
If you want to see the geometry of the now reverted ways, you can do that with overpass, for example https://overpass-turbo.eu/s/1Vc4 shows way/1339929485 as it was a couple of hours ago.

It looks like this has change might have broken some multipolygons (you can see them in the JOSM validator):

civil parishes
< -2235928 | Taney
< -2234306 | Stillorgan

Elsewhere these townlands were no longer valid multipolygons this morning:

< -4622479 | Rocheshill
< -4592163 | Ballinclea
< -2195324 | Mount Merrion
< (55675 rows)
---
> (55672 rows)

and these electoral districts:
< -12495187 | Stillorgan ED
< -12491366 | Dundrum ED (1911)

and these Anglican parishes:
< -10742937 | St. Peter's Parish
< -10709157 | Stillorgan Parish
< -8851833 | Stillorgan CofI Parish 1900
< -8811305 | Taney Parish 1900

See comment on changeset/159793833 - in the case of way/1334140977/history at least what would have made most sense to preserve the "railway=abandoned" tag on e.g. way/1334140978 , the track that is there now. The influence of the railway remains on what is there now; railway=abandoned communicates that to current data users.
In addition to that, it also makes sense to contribute "how things used to be" as actual railways to OpenHistoricalMap.

While way/1334140977/history was pretty obviously an abandoned railway based on the imagery, what would have made most sense here would have been to apply the railway=abandoned (or similar) tag to the extant track here way/1334140978 . Your changeset comment here is factually inaccurate here in that there IS still a trace of it in OSM. It didn't make sense for the other mapper to duplicate the way in OSM, but to just delete the duplicate without preserving the tags doesn't make sense.

OK, relation/14510149 is a valid relation again. https://overpass-turbo.eu/s/1V39 will show the relation at this point in time. Your version, which had all the duplication in it, was https://overpass-turbo.eu/s/1V38 . I'd expect that you'll want to add existing ways to the relation to do that, but you'll need to make sure that you don't duplicate any geometry of the current relation, which will break it as a multipolygon.

Hello,
Your last two edits have broken the Jersey National Park so that it is no longer a valid multipolygon. I suspect that the easiest way to fix it is if I roll back those two edits, you change again what needs to be changed, but then (and this is the important bit) you then check it with JOSM's validator to make sure that it's still a valid multipolygon.
Let me know if you'd like any help with any of that,
Best Regards,
Andy

Hello,
No idea how that happened - my recollection is that it's definitely an actual road!
Best Regards,
Andy

Hello c710,
What are you trying to test?
Whatever it is, please don't use the live OpenStreetMap server for testing. If you'd like help with whatever you are trying to do, please ask at https://community.openstreetmap.org/c/help-and-support/7/none .

Best Regards,
Andy

I've never been here, but the geometry that you've set for way/199340825#map=17/42.436709/12.409621 seems extremely unlikely. You're suggesting that there is no railway platform under the railway station roof, which would make it unlike any other railway station that I've ever visited.

@RedSkies please do reply to the above comment.

Before asking about the deletion of node/12294362769/history in changeset/158555897 I'm wondering how a traffic-sign=yes came to be added in a "removed features that no longer exist" changeset? Presumably it also added what is there now based on imagery?

Ah, I didn't realise that iD suggested different tagging for trunk and primary (although historically people have misinterpreted iD's "Not Specified" as "I need to set a value here").
The UK tagging of "green=trunk; white=primary" is widely accepted (see the reaction to https://lists.openstreetmap.org/pipermail/talk-gb/2021-September/thread.html#27754 , also mentioned at highway=trunk , osm.wiki/Roads_in_the_United_Kingdom ) isn't especially "odd", even though it doesn't match what some authorities might classify roads as - it's an easy to verify rule.

Also a kitchen shop on Blossom Street

The Esri imagery currently shows construction ongoing, but construction is basically complete. Despite that, it's still useful for locating features. The rest has been drawn by eye from roads and paths. More needs doing, especially when the basins are full (it'll only be possible to see what is where once they are - everything is just "undulating farmland").

Indeed - part of that discussion is happening at https://community.openstreetmap.org/t/is-this-mapping-historic-features/122139/7

Hello,
"bicycle=yes" is the default on highway=trunk roads in the UK (and most worldwide; it's only a few countries that have decided to use an odd definition for trunk).
I'm guessing you're working around a limitation of a router not designed to work in the UK?
Best Regards,
Andy

@GBAB - I suspect your changes to the Irish ones here was an accident! You might want to either change it back or liaise with the people who added that data to make sure that they are OK with the new tags.

OK - now done - see relation/192467#map=14/51.23258/-2.33116 and relation/9242266#map=15/51.23172/-2.32174 . Does that look OK now?

Hello,
I wonder if you can help. The relations relation/9242266#map=16/51.23216/-2.32587 and relation/192467#map=15/51.22876/-2.32977 seem to be used for different parts of the same cycleway link in Frome - I'm guessing that it's not supposed to be link that?
Best Regards,
Andy