• PersonalDataRepository

Since I first discovered WikiSoftware I've become more and more fascinated with figuring out a way to manage and share all my data from and to a multitude of platforms. I've had many half formed thoughts, most recently I've been influenced by the simple functionality of PyBlosxom (BlogSoftware inspired by Blosxom). Here's another attempt to describe what I have in my head, in the hopes that actual software will be inspired by a working description:

See also: WikiSpace, WikiAsPim, DigitalPalace, WikiToDrupal, InformationClients, AdamShand/2004-06-19

Requirements

• Should be file system based (for ease of management and replication).

  • See AdamShand/2004-06-19 for further thoughts on LabelsVsHierarchy

• The source file itself should be authoritive (eg. EXIF/IPTC for image MetaData rather then text or SQL data).

• Where additional MetaData is required (for searching/indexing/caching) it should always be possible to dynamically regenerate it from the source files.

• Source data should be available in as many different ways as possible (HTTP, DAV, IMAP, NNTP, LDAP etc).
• It should be possible to share portions of your data based on a global security policy.
• Ideally should support versioning (DAV(?), CVS, Subversion, etc).
• Items should be classifiable by with a global taxonomy, items should be able to belong to multiple categories.
• Security policy should allow access on a per-person or per-group basis.

Types of Information/Data

I've tried to determine the core types of data that I want to manage. Currently I manage all of the types of data with different systems with varying degress of sucess. It's hard to classify all the disparate data types ... here's a first pass:

• /writing (txt, wiki)
• /media (mp3, ogg, avi, mov, jpg, png)
• /email (maildir)
• /bookmarks (xbel?)
• /addresses (vcard)
• /calendar (ical)
• /code (.c, .pl, .py)
• /metadata (xml, sql whatever)

Other less well formed ideas:

• /reference (3rd party data, dictionary, thesaurus, cia factbook, gis data, wikipedia, etc)
• /genealogical (gedcom)

• /cache (SquidProxy cache for UniversalPersonalProxy type thing)

• /stickies (notes, unfinished writing etc. See Queue for interesting implementation)

Views

This isn't very well thought out yet.

HTTP

• /date
  • /2003/09/16
• /facet
  • /technology
    • /software
    • /hardware
  • /places
    • /usa
      • /oregon
    • /newzealand
• /rss

Import and Export

Importing / Editing

Should be able to get data from wherever I am to wherever it belongs.

• Email to anywhere in the (strip out attachments to /meda?)
• Filesystem (cp, mv, scp, rsync etc)

• WebDav (CVS, Subversion, etc?)

• Wiki

• XmlRpc (Blogger API etc)

• Ical client (Calendar.app, Mozilla etc)

Exporting

Ways to access the data:

• HTTP
• IMAP (more then just email could be readable via IMAP)
• NNTP
• LDAP
• XBEL (export all links as bookmark file?)
• XML/RSS/ATOM
• Ical client (Calendar.app, Mozilla etc)

Resources

• TwistedPython has pure PythonLanguage implementations of many protocols.

• OpenLdap has a shell backend for the data.

Questions / Thoughts

• Do I stick with the Blosxom model of heirarchy and use directories/links for categorization, or do I try and do something more abstract along the Drupal lines?
• How to get data from 2003/09/20 and in /technology/hardware?
• Could facets be a directory and symlink to the data?
• How to manage links?
• Can I just keep everything in CVS/Subversion? How do they handle links?

  • There is a similar solution for your idea if I have understood that well inside one of article of O'Really. Have a look at http://www.onlamp.com/pub/a/onlamp/2005/01/06/svn_homedir.html

  • SubVersion is able to handle links on 'NIX platform

• How does DAV versioning work?

• How much of what I want does the Web Crossing product do?

CategoryProgramming CategoryWriting