relative paths in IE XML / Storage Migration
Rosetta 5.2 has brought a new Storage Migration feature that enables institutions to restructure their permanent storage to fit changing requirements. However, the documentation currently warns that "Disconnecting legacy storage is not recommended as it may prevent the possibility of reverting to a previous IE version." (https://knowledge.exlibrisgroup.com/Rosetta/Product_Documentation/Version_5.2, Preservation Guide, page 185). As far as I can see, this limitation is due to the use of absolute paths in the IE XML files. The paths in older versions of these files cannot be updated without forging provenance information, which, of course, isn't an alternative. However, updating these paths would be the prerequisite for reverting to a previous IE version.
The documentation isn't clear about Rosetta's behaviour concerning older IE versions. These are the scenarios that I see:
- Rosetta might COPY older IE versions to a new storage to have the complete history stored on the new storage. The original copy on the legacy storage is kept to preserve Rosetta functionality concering version recovery.
- Rosetta might MOVE older IE versions to a new storage to have the complete history stored on the new storage. Reverting to older versions would not be possible anymore, but the legacy storage would be cleared of all files and could be removed. The complete provenance information would be kept in older versions of the IE XML, but with some invalid file paths.
- Rosetta might KEEP older IE versions on the legacy storage and write only new information to the new storage. Reverting to older IE versions would be possible, but the legacy storage could never ever be removed. You couldn't even clean up the mount points without losing access to older versions.
From my point of view, a possible way to go would be to change Rosetta's behaviour concerning file paths in the IE XML. Currently, absolute paths are used to point from the IE XML to the payload files/master images. However, using relative paths instead would make more sense here, because (at least on our storage. Comments?) IE XML files and payload files are always kept closely together anyway, so path complexity could be removed. Also, storage migrations would not make a path rewrite necessary, because the files can be addressed by the same relative path. For older IEs, that would mean that only versions created before the first storage migration cannot be reverted to. All subsequent versions would contain relative paths and could potentially be addressed even after a storage migration. For newer IEs (ingested after change to relative paths), that would mean that all versions can be addressed, even after storage migrations.
As this idea is somewhat storage specific (depending on storage layout, storage plugins etc.), I'd appreciate comments from other customers that see possible caveats. Context about SLUB's storage layout and our plans to redesign the storage can be found in the public SupportCase 00345262 "migrating permanent storage to new path [Rosetta 126.96.36.199]".