Memento Demos

Watch the movie that illustrates the power of the Memento protocol using the Memento Time Travel extension for the Chrome browser.

Try it out:

Install the Memento extension for Chrome.
Visit the continuously changing pages at LANL or Old Dominion University that we have set up for your enjoyment.
Select a date for time travel by clicking the black Memento extension icon: set the date to some value between 2009-10-12, when archiving of the page started, and now.
Right-click the page and select the "Get near ..." option from the Memento sub-menu. You end up at a prior version of the page, dating back to around the selected date. Notice that the images are prior versions too.
Don't forget to right-click some links at the bottom of the page in the same way. Some lead you into Wikipedia. But as you will notice, not to the current Wikipedia pages ... You are time traveling!
On a prior version of a page - the Memento extension icon is now red - right click and select the "Get at current time" option to return to the present.

For the Linked Data advocate who understands that data can't exist without versions:

Install the Memento extension for Chrome.
Select a date for time travel by clicking the black Memento extension icon: set the date to some value between 2007-09-04, the date of the first DBpedia version, and now.
Enter your favorite DBpedia subject URI in the browser's address bar, e.g. http://dbpedia.org/resource/DJ_Shadow.
Right-click the page, select the "Get near ..." option from the Memento sub-menu. Keep right-clicking links. You are time traveling!
On a prior version of a page - the Memento extension icon is now red - right click and select the "Get at current time" option to return to the present.

For the HTTP techie who wants to understand the Memento magic and start hacking away:

Install the Memento extension for Chrome.
Install the HTTP Headers extension for Chrome.
Now just repeat the above demos, and check out the HTTP headers used in Memento: Accept-Datetime, Memento-Datetime, Vary, Link, Location.

For those who want to understand the power of the resource versioning approach leveraged by Memento, this flip book movie is a fun illustration. This is how it was created:

For months, subtly changing "pictures of the day" were published that depict members of the Memento team posing next to a flip chart. The changes are in the shirts, the timestamps, and the BBC or CNN content superimposed on the flip charts.
At any moment in time, the current picture of the day is served from the following URIs:
- http://lanlsource.lanl.gov/pics/picoftheday.png for Los Alamos;
- http://odusource.cs.odu.edu/pics/picoftheday.png for Old Dominion.
On a daily basis, these images are archived using a transactional archive approach. As a result, old versions of these pictures become available at URIs like:
- Los Alamos picture originally served on March 20th 2010: http://mementoarchive.lanl.gov/ta/20100320180023/http://lanlsource.lanl.gov/pics/picoftheday.png;
- Old Dominion picture originally served on October 9th 2009: http://mementoarchive.cs.odu.edu/ta/20091009160013/http://odusource.cs.odu.edu/pics/picoftheday.png.
Using the following script that targets the URI of a picture of the day (say http://lanlsource.lanl.gov/pics/picoftheday.png) with incrementing datetime values for HTTP content negotiation and "follow your nose" HTTP navigation, all pictures of the day were gathered. In essence, the URI is asked about its past, and the URI gracefully responds by pointing at a TimeGate that knows about its past, and redirects to archived images that match the requested datetime values.

#!/usr/bin/python import os, sys, urllib2 from datetime import datetime, timedelta fmtstr = "%a, %d %b %Y %H:%M:%S GMT" img = "http://lanlsource.lanl.gov/pics/picoftheday.png" timegate = "http://mementoarchive.lanl.gov/store/ta/timegate/" first = datetime.strptime("Mon, 30 Sep 2009 19:00:00", fmtstr[:-4]) (now, td) = (datetime.now(), timedelta(1)) (savePath, typ) = ('/path/to/save', 'png') def get_memento(t): req = urllib2.Request(timegate + img) req.add_header('Accept-Datetime', t.strftime(fmtstr)) fh = urllib2.urlopen(req) data = fh.read() when = fh.headers.dict.get('memento-datetime', '') dt = datetime.strptime(when, fmtstr) if when else None fh.close() return (dt, data) def iterate(): (t, i, odt) = (first, 1, None) while t < now: (dt, data) = get_memento(t) if dt != odt: fn = os.path.join(savePath, "%04d.%s" % (i, typ)) fh = file(fn, 'w') fh.write(data) ; fh.close() odt = dt ; t += td ; i += 1 sys.stdout.write('.') ; sys.stdout.flush() iterate()
The gathered pictures were then concatenated to obtain this flip book movie.