Strategy to avoid duplicate (physical) FileEntries within DL

[7年前 に Steffen Schuler によって更新されました。]

Is there already a configuration/implementation out there that enables me to avoid duplicate file entries on the file system itself?
Maybe a hashbased solution? Maybe another FileSystemHook or a specific JCR implementation? Any idea is welcome : )

Currently, for multiple JournalArticles we generate according attachements.

JournalArticle v1.0 => DocumentLibraryFolder + DocumentLibraryEntries
JournalArticle v2.0 => DocumentLibraryFolder + DocumentLibraryEntries (copied from previous version)
JournalArticle v3.0 => DocumentLibraryFolder + DocumentLibraryEntries (copied from previous version)
JournalArticle v4.0 => DocumentLibraryFolder + DocumentLibraryEntries (copied from previous version)
JournalArticle v5.0 => DocumentLibraryFolder + DocumentLibraryEntries (copied from previous version)
...

If we have 30 versions of an JournalArticle, in the worst case we end up with 30 duplicates of fileentries!!! Some Attachments are 100MB large!

[7年前 に Olaf Kock によって更新されました。]

A hash-based implementation sounds good. Just be sure to also work on the DELETE semantics: Deleting a version should only delete the underlying file when there's no more entries referencing it.

And no, I'm not aware of such implementation to already exist.