(I am happy to write+contribute this feature, and am mostly interested in community interest in this feature.)
There is an S3 feature ("object lock") for compliance purposes that will keep arbitrarily many version of any object in a bucket.
When overwriting an object at a particular key, this will keep the prior version for the specified duration.
There are one in a trillion odds of two objects hashing to the same 128 bits, after storing 26 trillion objects, more or less.
If there is more than one version of an object, someone is doing something funny!
I think it might be useful to have a flag that one can set, to allow the S3 backend to check how many versions of each file exist before fetching, and then abort when the number is >1 for any file.
(I am happy to write+contribute this feature, and am mostly interested in community interest in this feature.)
There is an S3 feature ("object lock") for compliance purposes that will keep arbitrarily many version of any object in a bucket.
When overwriting an object at a particular key, this will keep the prior version for the specified duration.
There are one in a trillion odds of two objects hashing to the same 128 bits, after storing 26 trillion objects, more or less.
If there is more than one version of an object, someone is doing something funny!
I think it might be useful to have a flag that one can set, to allow the S3 backend to check how many versions of each file exist before fetching, and then abort when the number is >1 for any file.