OCFL on Append-Only-Filesystem

[je4]

If a append-only file system is used, which allows files to be added but not changed or deleted, Section 3.7 prevents OCFL from being used if more than one version is required.

Every OCFL Object MUST have an inventory file within the OCFL Object Root, corresponding to the state of the OCFL Object at the current version

Section 3.5.1 states, that the head defining the "current" version must be the latest version

head: The version directory name of the most recent version of the object. This MUST be the version directory name with the highest version number.

Therefore, the head can easily be calculated as the highest version number. This means, that there would be no loss of information, if Section 3.7 were changed to:

Every OCFL Object MUST have an inventory file within the OCFL Object Root, corresponding to the state of the OCFL Object at the current version OR an inventory file within the (latest) version directory.

This would allow new versions to be added without modifying the OCFL Object Root (except for adding a new folder or files).

[neilsjefferies]

@je4 Editors would like a bit more information about your Use-Case (e.g. which file system are you using). Your approach seems reasonable but introduces some issues w.r.t compatibility with the current spec. so we need to be careful.

[je4]

@je4 Editors would like a bit more information about your Use-Case (e.g. which file system are you using). Your approach seems reasonable but introduces some issues w.r.t compatibility with the current spec. so we need to be careful.

there are two cases, i would like to use:

1. 1. Our S3 Storage can be configured for not allowing modifications on bucket data except creation of new files. this would prevent any type of accidental deletion or modification
2. If you want to use an archive-format for the ocfl object, you can just add new content (i.e. for tar just append at the end...)

i can see the problem, if ocfl libraries want to get an inventory directly in the object root. but for version 2.x, the whole stuff will be broken anyway and maybe, there will be something like a compatibility mode, which would enforce the inventory within the storage root.