How does the Sync Engine determine if a file has been modified?

[Elkeid-me]

Sometimes, I accidentally modify a note. Even if I undo the changes in time, it seems that the timestamp still gets updated, and the Sync Engine uploads the file during synchronization (even though the file content hasn't changed).

Is it possible to store file hashes to prevent this unnecessary synchronization?

[hesprs]

Sync Engine uses the last modified time + file size to determine the modification. Changing any of the two makes the file "modified".

Storing file hashes would be more accurate, but it will need to hash the entire vault each sync. Which is a performance nightmare if the user has thousands of files or some extremely large files.

[Elkeid-me]

Sync Engine uses the last modified time + file size to determine the modification. Changing any of the two makes the file "modified".

Storing file hashes would be more accurate, but it will need to hash the entire vault each sync. Which is a performance nightmare if the user has thousands of files or some extremely large files.

Is it possible to use the last modified time + file size to filter for "suspected modified files", and then only hash those specific files (instead of hashing the entire vault)?

[hesprs]

Then Sync Engine still needs an additional hash database, perform a full hash at first upload, and update hashes incrementally.

Also, modified time change is meaningful in some cases. For example, #159 would like to sync creation and modification time.

Hash-based change detection in plugin core is adding complexity with minimal gain. But it could be a nice idea as a Sync Engine module.

[hesprs]

On remote side, Sync Engine uses whatever best marker it can find to identify file modification, which is often the Etag of a file. The implementation of Etag varies across servers, but it is often hash-like so redundant downloads can be skipped.