[tahoe-dev] wanted: a permanent copy of everything I've ever looked at through my web browser

[Aleksandr Milewski]

On 2/16/10 6:10 PM, Zooko O'Whielacronx wrote:
 >
 > But, how do I hook up the Tahoe-LAFS storage grid to my web browser?

For integration with the browser, and something that's already open, you 
should look at Weave.

http://mozillalabs.com/weave/

It's designed to keep your bookmarks, browser history, and tabs in sync 
across multiple instances of the browser, has decent crypto properties 
and is open and documented.

If you really want to archive all the content, you would want to set up 
a client that pulled all your history, scraped it off the web, and 
stuffed it into Tahoe.

Assuming that scraper was running most of the time, you could use 
Mozilla's Weave servers (which I maintain, btw).

For various reasons, including control over how long you keep history 
available, you may also wish to run your own server, which isn't too bad 
if you aren't designing for millions of users. (We are, which has made 
the system a little more complex)

> Goals:
>
> 1. Storage of everything that I ever look at in non-lossy form so that
> it can be understood by future versions of this tool (ser below) and
> other tools and my future self when I ascend into a cyborg
> archaeologist.

See above.

> (Goal #1 should be implemented and deployed before the rest of them so
> that I stop forgetting things.)
>
> 2. Browseable in chronological form -- what files I looked at when.

Weave timestamps records, yes.

> 3. Browseable in hyperlinked form -- when I was looking at this file
> and clicked on this link in it then I saw that file.

You want a click-track? That's a bunch of work deep in the browser. 
Inference from chronology is easier, and would be enough for me.

> 4. Searchable.

All the caveats for searching ciphertext apply (you can't). Having said 
that, it's worth exploring the browser history settings in Firefox to 
see if you can make the time horizon infinite, or close enough to it 
that you don't care. Places (where the history is stored) is a sqllite 
db, so it should be able to handle pretty good sized datasets. This will 
likely necessitate running your own Weave server, however, as we don't 
sync really old stuff to ours. (30 days sticks in my mind, but I could 
be wrong)

-Zandr