Notably becaue when resolving them (I detect validation errors in edits from others: I revalidate then to fix holes and this makes edits elsewhere, notably by countries)
Initially it was only in France overseas.

it was VERDY small, but interrupted by edit conflicts in the Netherlands (Europe).
It was intended in a small part of Saint-Martin.
Resolving edit conflicts can make edits much larger than expected

Sauf que "ce qui est périmé" était impossible à évoluer. Et que partout autour le reste méritait ces corrections.
En soit je n'ai rien cassé puisqu'il y avait un parking (tracé un peu n'importe où en superposition d'autres objets, ce qui était l'objet essentiel de la correction).
Il fallait bien m'appuyer sur quelquechose de vérifiable: je n'ai PAS supprimé ton parking, ni son sens unique, même s'il y manquait le détail des allées (impossible à déterminer), et au vu des autres travaux autour, rien ne permettait de dire que c'était déjà terminé à l'époque. J'attendais seulement que ce détail soit réellement visible, sinon impossible de positionner les autres petits éléments: par exemple les emplacements de poubelles que tu avais mis et qui étaient aussi incorrects et que j'ai corrigé car les containers sont parfaitement visibles maintenant.
Je n'ai pas fait ça "en aveugle". Pour le détail précis des allées de stationnement, c'était mineur et cela pouvait attendre, il n'y avait pas d'urgence.
C'est sur ce genre de points qu'ensuite les locaux peuvent affiner un peu plus... quand l'imagerie devient réellement disponible pour le faire (ce qui n'était pas encore le cas à l'époque).

Bref je n'ai pas fait ce que tu disais : "tout bousiller", c'était propre

Somme toute admet que je n'ai rien cassé gravement" et que c'est grace à ce ménage que la prise en compte de l'actuel a été facile et rapide à faire (même de façon précise, sans bâcler)

Et j'y retourne aussi régulièrement (pas spéciualement ce quartier, mais Rennes et sa périphérie je connais bien)

Et puis tu as déménagé de la zone, moi aussi j'y ai vécu.

la "fraicheur" n'est pas évidente à évaluer. Les dates des imageries ne signifient pas grand chose. C'était dans une zone assez fouille et très peu précise, où tout se malangeait, j'ai fait le tri en fouillant justement plusieurs imageries.
Tel que c'était tracé cela ressemblait surtout à un tracé très grossier, fait à la va vite. Note que j'ai été aussi précis que possible pour que la suite reste claire, ça n'a pas pris ensuite longtemps de modifier un peu le tracé avec l'imagerie plus récente, mais note bien que c'était il y a bien des mois !
Quand on commence à modifier des zones, il serait bon de faire le méange autour et remettre un peu plus de précision, sinon ça devient des spaghettis avec des traits dans tous les sens, et rien n'indique ce qui est supposé être plus récent ou meilleur (out est mélangé justement).

un an après...

Also note that I'm NOT alone to edit in this area; I fix many repeated errors made by jeje23 (incorrect islets added with reversed costlines, broken relations forgotten in most of the time...).
Unlike you I refresh my data frequently before ALL submissions, and I run again the JOSM validator. I do not need external outdated QA tools

Given the tone of your response, it's evident you cannnot collaborate with any one on OSM and you can harass them just like me at any time. I will complain tyo admins for your attitude. You are not patient and have not understood anything about how these QA tools works, and the fact they need time to be in sync: all external QA tools will frequently report "errors" for recent edits they've still not fully processed. It's evident that you misuse them and give them an incorrect level of trust.
I repeat: there's NO error, and you've not given ANY evidence that there was really "errors". You have NOt tried to load the data or see them by trying to load the data for example in iD, jsut to make sure that these reports are not obsolete.
Unlike you, I've checked multiple times my edits before and after submission and found absolutely NOTHING.
And given the kind of edits you make everywhere and randomly in the world, without any meaningful comments (always the same), it's eveident that you work only alone from you chair and have never tried to cooperate with any one, you just use your own "knowledge" and bad assumptions, and have not read any doc about the QA tools you use blindly: all what these tools say is: "please check here if there's a real problem".
OSM Inspector is correct but in France here, we've all decided to ignore its too many false positives and we use Osmose instead (which is more complete, more accurate, and in sync after 3 days at most, where OSM Inspector needs weeks or months, or never cleans its old reports in some areas.

And I speak about DATA, that you have NOT checked at all, not even tried.
You would have seen that the reports from the QA tools are wrong because they were based on different old data and are not relevant to my recent edits.
Sorry but you don't know how to read these QA reports (notably the dates and given IDs).
I'm not trapped, but YOU are: you incorrectly interpret what these QA tools report and read them superficicially!
All existing external QA tools ARE slow and never give reports accurately before at least several days: the reports are just indications and need further checks.
Compare what these reports say with what you see in the iD editor: things are different including the object id's, the geometries (notably the total way length when the way was split), the tags...
As well as the attached meta data (notably the changeset id for each object: if they don't match the changeset id shown in the QA tool, the report of the QA tool is not relevant, and will be cleaned by them automatically on their next scanning pass which never occurs instantly but always once every few days)
Read the docs about these QA tools: this is always clear that they are not showing "errors", but warnings about possible errors if they are not checked manually.
Unfortunately the reports given by OSM inspector are not very easy to intepret.
I use other QA tools which are more precise (and for this location in France, OSM Inspector is almost never used by French contributors that prefer Osmose)
You also affirmed that I disabled and did not want to use the JOSM validator, and I can affirm you that I always use it strictly and, I never ignore it, it is always enabled by default. I also perform lot of other checks, even if you think the contrary, you're not behind my back to see it, but given that you do not even check what is really in the db by trying to load the data yourself, your claims are invalid.
You'll react the same way with any one else editing data anytwhere in the world if you interpret the QA tools reports the way you do now. For now you are just creating troubles. You need to learn how to use and interpret these external QA tools, and work really in cooperation with others editing in the same area as you, because you'll just harass them needlessly just like you do here.
So keep quiet, and learn.

"The error was yours. JOSM also should have warned you before upload. And you either have deactivated the JOSM checker or ignored the warning. But I'm pretty sure that JOSM is also wrong, right? The tools are working absolutely correct."

I completely deny *ALL* these false affirmations. I've not disabled these things in JOSM. Stop this polemic. You just affirms that the external QA tools operate instantly, and this is wrong. It takes them several days to really see what happens, even if they've loaded the data after a few minutes, they don't process the data instantly, and keep their reports online. Zones are processed by them not in the minute, but only once every few days. As well tiles are not rendered instantly (especially for low zoom levels), and even if they are marked for being regenreated, it takes time (more than one month for coastlines in renderers).
So you're just impatient. If you want faster results, donate for hardware to those that really put these tools online.
But I can affirm that you've not even tried to load these areas by just trying to edit them in iD: you would have seen that what is reported is not even there in the database and that the reported "errors" do not even exist and are base on a mix of old data and new data has still not been processed by these tools.

If you're patient enough, the QA tools will see the effective changes and will no longer report the alledged "errors", which were there since long before my recent changes but are not caused by my changes that have already solved many of them.

The way/686907210 was exactly created when I splitted an existing way (which was not duplicated) in two parts to delimit a new bay.
But there was another older way (created 6 months ago, not by me, by someone redrawing a part of the coastline without connecting it to any relation) that I deleted at the same time (the way reported by OsmInspector which is incorrect), because it was already a duplicate before my change there.

yes that way was created from a split of an existing one, and it was alreadfy part of relations.
The way the was spurious was createrd 6 months ago and referneced by no relatioçn at all.
You're wrong. Tools are good only after some times to see what's been forgotten, but I confirm that OsmInspector is wrong here: the Ids indicated either don't exist anymore or are not the way they are shown because it was split in several parts and one part merged with another, and OsmInspector still does not have it.
I've have reverified, everything is OK.

I also confirm that EVERYTHING that OsmInspector still reports is wrong. Thery are all based on old datas, and not at all about what I did in the last few days.
OsmInspector is useful only to find things forgotten for more than 1 week ago, in much more frequently edited areas in Europe (but in this area it is even late by several months!)

Also take what "OSM inspector" affirms with care: look at the dates of the data: these dates in its reports are even BEFORE ALL I did there !

I confirm: the error you see is a duplicate way that was created more than 6 months ago, and incorrectly tagged as a coastline, but never used by the relations that were already there.
So it's definitely not me that created the duplicate.

OSM inspector did not detect that a geometry was redefined (the two ways were alreayd merged by me into one, it keeps a reference to a deleted one: the duplicate was there BEFORE my changes, I have explicitly removed these duplicates, which were cased by old baselines incorrectly tagged, or old bays incorrectlmy drawn with overlaps)