This sounds reasonable, since nothing bad happened here, hence it's not "Lacuna per se".

My automatic script did a weekly "full update":
– full manifest update (nothing new, nothing changed) -> manifest updated
– full verification of everything (nothing new, nothing changed) -> manifest updated
– download "per ID" (i.e. explicitly for each "textual" ID; as expected: nothing got actually downloaded) -> manifest unchanged
The manifest is still there, having slightly changed during each of the 1st 2 steps, but not during any of the download commands (including when having been called specifically for the [unchanged] "lacuna_a_scifi_noir_adventure" explicitly).

Thank you for all the help. I'll make sure to give Avalon some extra scrutiny.

Yes, my manifest isn't finished rebuilding yet but looking at the game information file on GOG, I can see that Fall of Avalon has a DLC with a changelog entry which from a quick look through my old manifest seems to be entirely unique, and that seems likely to cause problems ( the changelogs require special handling because as text blobs they are allowed to have character sequences in them that would usually indicate the AttrDict was messed up ) .

ETA - The really weird thing is that it has a changelog entry in it's DLC which seems to be unique but that changelog entry is empty anyway. Truly a rebel without a cause.

The resume manifest shouldn't be effected because it's a special file that just contains a list of games that have been seen but not yet processed, it's there so full updates know where to resume if interrupted and so that the deeper inspection that clears the new / updated flag in your library won't permanently render it unknowable what games for updated / are new if a normal check gets interrupted.

So the issue will only affect those who own that respective DLC, then?

So I assume as long as we manually clear the blue dot from Avalon, we should be safe to use GOGrepoc standard update?

[Update 01]
Upon second look, I did update tainted_grail_the_fall_of_avalon on Saturday (2026-06-06) and saw no issues. There is no changelog in my library or my info file though.

So I am either using an older script that is bug-free (Is there a version check command or something?), or maybe it's a CDN/Regional thing / language thing?

Maybe not. I was wrong about the cause. Turns out the extra changelog entry doesn't matter because never thinking about it happening means it never got processed and got tossed into the void eternal, so that's not the culprit, just a coincidence. I've finished rebuilding the manifest now so I can now as gogrepoc to tell me what the hell it thinks is triggering the clenaup.

ETA - Appears the issue is not happening off my rebuilt manifest. That's really annoying.

ETA2 - I need to do some maintenance stuff to prevent it redownloading a bunch of installers unnecessarily ( MD5 mismatch but only internal to the manifest, not necessarily a real MD5 mismatch ),., but I'm going to do that then a full sequence just to verify the issue doesn't seem to occur anymore.

ETA3 - Will likely be a while before the next update, since I have to verify over a TB of files to prevent having to redownload them ( need to tell the manifest that the MD5 of the files matches the final MD5 in the manifest even though the MD5 manifest has chanced ( because I had to rollback my manifest a few months )). I'd expect that to take at least 4 hours and I'll hopefully be in bed by that point. I probably need to add an option to -clean to handle this scenario automatically since the manual way isn't super obvious ( after a manifest rebuild from an old manifest, you need to verify all files before doing a clean or download not using -permissivechangeclear which could clear some files even if they shouldn't be ) .

ETA4 - Took 6 hours for the verification. Need to update the manifest with the new updates from today, then do the downloads. Then I'll back up the manifest and do a full update again just to verify that the issue isn't happening. Very annoying to encounter such a major error but be unable to replicate it ( so far )

ETA5 - I think the problem may have been Outer Worlds Spacer's Choice because there's a weird inconsistency with the behaviour of my existing folders on Extra/Linux/Mac compared to what happened with Windows immediately before the problem happened but it doesn't seem to be causing an current problems so far but if it's fixed and I can't replicate it there's not much I can do.

No, probably wasn't Epic –there's been a german language thread about this, and one person filed a complaint to GOG about removing the original, replacing it by a modification as the main version. This incident was when they changed it back, bringing the modification from version 6.3 to 6.66 at the same time.

Welp, I've got my manifest back up to date and the problem hasn't recurred, which means it's probably a unique problem from having a manifest from some point after March and updating a particular (as yet unknown) game to a relatively recent (within the last month ) update. I'm going to run through a full update from my current manifest just to verify that things are stable now but unless someone has a pre and post error manifest I don't think there's anything I can do ( maybe automatically make a special backup outside the usual rotation before repairing AttrDict ? )

No luck with reproduction, everything went off without a hitch. I guess all I can do is recommend caution and making a manifest backup if you're updating a manifest from after March 1 but before early June 8th and to contact me if you have the problem. I've gone over the detection code and I can't see anything that could be causing problems so whatever the edge case is has to be very weird.