search for single repository

Posted by Alexander Deliyannis
Nov 1, 2008 at 08:27 AM

I would agree with Steve Z. that a single repository for everything is for now unrealistic—and, as far as my own needs are concerned, probably for the future as well.

For example, I have a large quantity of project-related photographic material that would bloat any database file to the point of uselessness. I only recently delved into movie files, and already my library of TED (http://www.ted.com) lectures representing valuable references, has grown to several gigabytes. So I would suggest that the file system remains unchallenged as the repository infrastructure.

Focusing on the large percentage (in terms of discrete items rather than volume) of information that many of us have to deal with, I would suggest that much of what has been mentioned is capable enough, as long as it can incorporate links to items held in external repositories, be it the file system or other dedicated applications supporting such links to their content, e.g. Evernote, Surfulater, UltraRecall, Zoot etc. In this way, rather than exporting/importing information around, one can simply link to where that information is currently held. (Mind you, a long-term discipline in file management is required so that such links don’t become redundant overtime).

An additional reason for this is that the near future may easily bring more kinds of information items. E-mails, websites and files have been mentioned, but how about RSS? Omea Pro incorporated RSS as a repository and the first time I read about it I thought it was of marginal importance, but now RSS constitutes an important part of my reference information. OPML files grouping several RSS feeds together are now widespread and dynamically linking to them is preferable over importing. And, I’m sure there’s more to come.

In fact, I can’t stress the superiority of linking over importing enough: as flexible as all-in-one applications may be, dedicated applications usually do their job better. For example, even if one imports emails to a single mega-repository, one will probably keep sending e-mails from the original e-mail application. Another example are websites: with dynamic Flash-driven sites almost becoming the norm, reliably grabbing information from the web is becoming more and more difficult for general-purpose applications.

Last but not least, it’s incredibly easy to end up with multiple outdated copies of one’s data. I see “Outlook address book importing” in just about every software imaginable but, overtime, this may be useless if the imported data can’t be synced to the original. But then, if it needs to be synced, why not simply link to the original?

Cheers
Alexander