vrypan|net|log (archived)

Lately, I’ve been testing services like dropbox, sugarsync, etc. The reason is bucket3:
1. I’d like to have a “local” copy of my content and be able to add or update it,
2. then have it automatically synced to my server,
3. where bucket3 would notice the changes and update my blog.
4. I would also like to have content added on the server side, by other means (ex. a script downloading my twitter updates daily), synced back to my laptop.

In order for this to work, first of all, I need a file syncing service. Most of the above mentioned are good in syncing files -even though Linux CLI support is very bad, and my server is a Linux box. So they are on the right track.

dropbox has a nice feature: they have an RSS feed of all the changes.

Then it struck me. We need an open source file syncing service that would use RSS to describe the changes on the “main” file repo, and then use rssCloud to notify the rest of the repos.

It would work like this:
1. A local repo knows when a new file is added or modified and pushes this change to the main repo.
2. The main repo updates its changes.xml rss feed and uses rssCloud to notify all local repos of changes.
3. Local repos pull the changes and everything is in sync.

Of course, it might me as easy to set the whole thing as a decentralized system, where no “central” or “main” repository exists, and all repositories notify the rest of any changes.

An extra bonus: it would be very easy for developers to implement action hooks, the github style: for example, I’d like to run a couple of bucket3 scripts when new content is added or updated.

And the whole mechanism is based on open standards, widespread knowledge and experience with RSS, aggregators and tools like rsync making it easy to can have different implementations per platform of programming language or needs.