Time to say farewell to ‘Files-in-Folders’?

(No Time To Wait 8)

Peter Bubestinger-Steindl
(Peter@ArkThis.com)

2024-09

Digital without files and folders?! 😱️

https://nymag.com/intelligencer/2014/11/white-people-rioting-for-no-reason.html

Imagine being able to:

  • seamlessly annotate any data. Anywhere.
    (key/value + relationships)
  • basic and interoperable
    (as moving and renaming files/folders)
  • copy/move/merge any data from anywhere
    (without having to worry about file/folder naming and structure

I think I’ve found a way to…

“Go LEGO” on most our annotation needs

http://playmobilvslego.tumblr.com/post/103112937090/when-lego-and-playmobil-become-pieces-of-art

I do professional “Import/Export” business.

  • Of digital data.
  • Like all the time.
  • And it’s getting less funny.
  • And extremely expensive.
  • And it never seems to stabilize anymore.

What’s wrong with how things are?

(Dakota Nelson, DEFCON 25) https://youtu.be/uPbDySi-p2w

Sneak Peak into “the Future”

We can use ‘old’ (=stable) stuff now.

Extending Files and Folders (xattrs)

  • Get “key/value” on any “data object”.
  • Use 0-Byte files as “DB entry”.
  • Use this to depict relationships/links, too.
  • Right-click-edit-metadata as default.

Extending Files and Folders (xattrs)

Filesystem as Database?

Persistency/reliability concerns?

Anyone still worried to use more than 8 characters in a filename? Or Unicode?

Additional Workarounds

  • export/import xattrs as plaintext
  • tar
  • file a bug report
  • upstream patch
  • …
  • I really don’t think this is a show-stopper.

Finally, after decades…

Give filename and path the place they deserve.

  • Real key/value metadata fields.
    (Instead of being a special technical handle, with all known compromises)
  • Why not “0 or more” names/paths?
    per “file” (Data Object)?

“Naming” may even become optional?

If a ‘data’ has no name or path, it shall also exist.

  • As data (+ID) in a storage “pool” (or folder)
  • You just need something to remember it by.
  • Just press “save”.
  • And use auto-existing metadata (date, title, context, etc)
  • And add more info later as you like.

A quasi-unique-enough ID

  • For anything: meta and data alike.
  • Being able to address & use “catalog entries” and “files” alike.
  • So for copy/move/merging data collections, filename/path does not matter anymore.
  • Maybe even use “Collision Friendly IDs”?

Generic Meta+Data Building Blocks.

Once you have UIs for basic key/value/relationships, you immediately feel how suddenly things become very easy.

And very interoperable.

(comparable to XML/JSON instead of binary formats)

If Taken Seriously:

…a step-in replacement for?

  • embedded metadata (and format dependencies)
  • annotation spreadsheet
  • relational database
  • sidecar metadata files
  • …

What’s missing yet?

  • Proper User Interfaces (UIs) (Commandline + GUI)
  • and the mere will to “go from 8.3 to long Unicode” once more 😉️
  • to build a smooth computing environment supporting this.

This may provide us with:

  • A blueprint for a generic storage + catalog engine.
  • That is cloud/big-data compatible (+it scales!)
  • An inter-institution storage/data network option. (Wanna use/refer to anyone’s “data”? That’s really easy now.)”

So memory institutions may:

  • Stabilize their data handling/storage conundrums.
  • In a generic way.
    (for any collection types)
  • From now on refer to meta+data equally.
    (As filesystem “objects” (by ID))
  • Get out of the neverending-migration-update-import-export-exhaustion-madness™

Thank you.

Peter Bubestinger-Steindl
(Peter@ArkThis.com)

ArkThis AV-RD