class Ifconfig
  RX_REGEX = /^\s+RX packets:(\d+)/
  TX_REGEX = /^\s+TX packets:(\d+)/
  COLLISIONS_REGEX = /^\s+collisions:(\d+)/

  def collect samples = 120
    collected = []
    samples.times do
      collected < < row.join(',')
      sleep 1
    end
    header_row.join(',') + "\n" + collected.join("\n")
  end

  private
    def header_row
      ['date', 'rx packets', 'tx packets', 'collisions']
    end

    def row device = 'eth0'
      output = `/sbin/ifconfig #{device}`
      rx = output.match(RX_REGEX)[1]
      tx = output.match(TX_REGEX)[1]
      collisions = output.match(COLLISIONS_REGEX)[1]
      [Time.now.strftime('%Y-%m-%d %H:%M'), rx, tx, collisions]
    end
end

use it like this:

# collect data
require 'ifconfig'
ifconfig = Ifconfig.new
data = ifconfig.collect # get a drink while you wait

# stuff it into swivel
require 'swivel'
swivel = Swivel::Connection.new
data_set = swivel.upload :name => 'spork: ifconfig eth0', :data => data

# create a graph of RX packets over time in swivel
graph = swivel.create_graph :x_axis_column => data_set.data_columns[0].id,
  :columns => data_set.data_columns[1].id, :graph_type => 'LineGraph'
graph.save!

NOTE: wordpress kills my backslash-n newline characters. but you know, they’re there. copy/paste diligently.

[huned@spork swvl]$ sudo gem install swivel
Successfully installed swivel-0.0.8
Installing ri documentation for swivel-0.0.8...
Installing RDoc documentation for swivel-0.0.8...

it’s under moderate development still, so best to grab it from svn (read how) until the swivel api is done f’real.

validates_uniqueness_of login_token # triggers for nil. bad!

fortunately, those guys that make the rails go, they’re smart. you can change this default behavior and all will work well. simply set the allow_nil option to true

validates_uniqueness_of login_token,
   :allow_nil => true # doesn’t trigger for nil. good!

this is all in the activerecord doc, but it’s hard to find.

DEFAULT_VALIDATION_OPTIONS = { :on => :save, :allow_nil => false, :message => nil }

go forth and validate.

1. create the directory which you’ll import. e.g., for an addressbook app. make sure you include a trunk/ subdirectory.

   mkdir -p addressbook/trunk

   if you have any code, copy it into addressbook/trunk/.

2. now import that into your repository (which in this case is hosted on a machine named build). thusly for the addressbook app:

   svn import addressbook svn://build/addressbook

   this will slurp everything in the addressbook/ directory into svn.

3. let’s check out the addressbook app from svn to make sure it all imported successfully.

   svn co svn://build/addressbook/trunk

   when it’s done, you should have a trunk/ directory around. inspect the contents of trunk/ to ensure it contains everything you expect. it should.

4. at this point, you can remove all the original stuff and begin using your svn’d stuff.

   rm -rf addressbook

   mv trunk/ addressbook/

   .

for more information, please check the more authoritative resources are svn help import and the svn manual.

how would you do this in your web app if you felt so inclined? check the user agent in your code. if it’s a browser, don’t load.

(for now, please ignore the colors — i’ll change them soon.)

so, a couple few interesting observations.

• the obvious thing — why are there way more cd sales? one possible explanation is that it’s a marketing victory. remember those old BMG and columbia house deals where you could buy 30 CDs at a time for some obscenely low price? another, more interesting, explanation is that the cd players became commodity technology fairly quickly which made it widely available to more people, which means the market who consumes cds increased. i have no data on this, but it seems to make sense — electronics become commodities pretty quickly these days.
• the intersection of the tail of vinyl with the head of cds is approximately near the peak of cassette tapes. i wonder if this is a characteristic of technology obsolescence when it comes to music distribution technology.
• the approximate death of vinyl’s sales coincides neatly with the intersection of the tail of cassettes and the head of cds. same goes for the intersection of vinyl and cassettes — it’s approximately the start for compact discs. curious. again, is this a recurring characteristic?
• unit sales of ipods seem to fit in crisply with what we expect after looking at the previous technologies’ osolescence curves. the sales of ipods start at approximately the same time that sales of compact discs begin to slow. not a surprise, but it’s nice to see that actual disparate data sets confirm the regularity of obsolescence cycles. also note the positioning of the intersections. this is telling.

(matt pointed out that wikipedia has a good chunk of itunes music store data. i need to get this into swivel. thanks matt.)

what do you think? what other technology trends can we track for fun (and profit)? i could look at this graph for hours, but i need to make swivel better for a while.

credit for this goes to the swivel community:

seema created the original graph that caught my attention from from data uploaded by c.a.joyce. the data is sourced the world almanac and the riaa, with additional data on ipod unit sales from adpowers

my rationale for the animated gif progress bar is that it’s cheap to implement, reliable, and feeds the user back that “something is happening”, which is the primary goal of the progress bar. at kickapps, sadly, i failed to influence the decisionmakers into doing what i believed was the right thing. we ended up going the way of the ajax progress bar, which ended up being all kind of broken.

i just bought an airline ticket from orbitz. what kind of progress indicator do they use? the gif. here it is:

it’s not the best progress bar in the world, but it gets the job done. i personally like those horizontal infinite barber-shop pole kind of progress bars, which you can see over at mentalized. here’s an example.

other advantages of a fake-y animated gif progress bar? well, if you work for a startup in a crowded social-media space, you’re working with limited resources (money, people, time to acquisition, quickly diminishing barriers to entry). the solution is clear: pick the cheapest good-enough solution that gets the job done. you’re not winning any points for innovating on progress bar implementations, that’ s for damn sure. spend your time somewhere else, where it matters.

but today, i spent a little longer. i always wondered what the “authority” select-box did. how does technorati measure authority? i tried it figure it out.

• i performed a search – http://technorati.com/search/kickapps
• set authority select-box to “a lot of authority”
• i saw my personal blog excluded from the results. this is funny because i actually am an authority on kickapps. i work there, i’ve been there since the early days, and i’ve written a couple relevant blog posts on the topic. so , why’d i get filtered?
• also note that i saw the kickapps corporate blog excluded from the results. the kickapps corporate blog is the authoritative blog for kickapps. it should definitely be in the results when i select “a lot of authority”.

but wait, what’s this? technorati displays the number of blogs linking to each item in the search results list. it appears that the “authority” filter is simply a (perhaps modified) link counting strategy. blogs with “lots of authority” have lots of other blogs linking to them, while blogs with “little authority” don’t have many incoming links (like my personal blog and the kickapps corporate blog).

that’s a fine strategy for determining some relevance of a blog, but it’s certainly far removed from a legitimate measure of authority of a blog. silly technorati, you are misleading your audience and undermining your own credibility.

it would be useful if technorati renamed this authority select-box to “incoming links” or “link popularity” with options ["many", "some", "few"], or something to that effect.

one of my users was being (unusually) idiotic and forgot his password. and he couldn’t use the forgot password functionality because i never configured email. and, i still don’t want to configure email.

so, i had to manually change his password using the mysql command line. here’s the sql update statement. run it and ask your user to log in and change his password to something more sensible. execute this statement as your mysql root user:

update user set user_password = md5(concat(user_id,’-',md5(‘new_password’))) where user_name=someone

i should have fixed this out a while ago, but i had been procrastinating.