Recovery issue after using Time Machine

I had a data synchronization issue with Scrivener a few months back and have been trying to recover a usable copy via Time Machine. As far as I can tell what I’ve recovered is a usable copy but it seems to get corrupted when I open it with Scrivener.
It’s possible that what I recovered only appears to be complete (due to the way Time Machine does backups of open projects) but for the following reason I do not believe that is the case:
When I show Package Contents and go into the Docs folder there are the correct amount of files and all of the Modified Dates make sense.

If I then open the project in Scrivener- the modified dates change and project becomes corrupted.

I’m very grateful for any help and comments.
Thank you

Are you using the actual Time Machine interface (Space, the Final Frontier, and all that), or are you just poking around on the disk and dragging stuff out? TM was never designed to accommodate the latter usage. You should always open the Time Machine program itself, and navigate the history browser within it to find copies of your older projects, and then use the Restore button to retrieve them. If this results in a collision (restoring a file in the same location as a current copy with the same name) you’ll be asked if you want to replace the current version with the backup, cancel, or restore it alongside and rename the current version to avoid conflicts.

This was not done using the Time Machine interface. This was due to the fact the open project created permissions issues in Time Machine. I understand this leads to the possibility of a problem and that’s why I went into all the details about the modified dates in the files and the number of files in my copy. As far as I can tell the project I recovered appears to have the relevant files but when I open it with Scrivener it becomes corrupted.

Hmm, you might want to tap Apple support on this one, as it sounds like you’ve got some OS related issues they could help you resolve, especially with the permissions problem in Time Machine, that sounds kind of scary. I’d want to get that resolved as soon as possible so that you know you are getting good backups.

As for the project you are trying to open—well there is a lot of difference between what looks like and what is right. It kind of sounds like you’ve pieced it together from various places by using the modification date. In theory that should work, but if these pieces are coming raw from Time Machine—well I have no experience trying to do that so I’m not sure what the implications are. Whatever happened, it doesn’t sounds like everything is assembled the way it needs to be for Scrivener to recognise it, even if it all looks right. It could be that the pieces themselves are internally corrupt, if they were mismatched and incorrectly altered. Do you not have any automatic, or whole .zipped project backups stored anywhere? Even if they are older, you might be able to take an older one and by opening the RTFs from the damaged project, piece things together in Scrivener rather than trying to use the Finder.

I was talking to a backup expert who had this to say:

Time Machine uses hard links to store its backups of your files. It looks like Time Machine is storing all your files for each backup, but in reality all the unchanged files are just hard links to the same underlying file data as the previous backup. It’s a Unix trick and it works fine. Time Machine doesn’t use aliases. If you copy a file from your Time Machine backup to your desktop, you get the file, not an alias.

That is correct, that it uses hard links is exactly the problem. If I recall correctly one of your first experiments was to open the .scriv file on the Time Machine drive itself—and something went sour when you did that. Because a project file on a Time Machine drive is actually mostly just hard links with the few files that did happen to change in the past hour, what you essentially did was open up dozens of copies of the project at once—and when that fouled up—dozens of copies were corrupted instantly because they are all sharing many of the same internal resources—but not all of them. So parts will change at the file level, losing synchronisation with other parts that didn’t change and so on. You end up with 12 projects all slightly damaged because they no longer are internally consistent.

I’m speculating, but that is what it sounds like; and if that is what happened you might have an entire lineage of backups for this project in a damaged state: hence the problems when trying to open them after restoring them.

Have you tried creating a new blank project and importing one of the copies you are trying to open? Use [b]File/Import/Scrivener Project...[/b] that can sometimes salvage a lot of data out of a project that will not otherwise open.