random code

Ugly Code, Part 1

rjp — 2005-11-16T14:32:27+00:00

Inspired by Monitoring network traffic with Ruby and Pcap . The pcaplet library for Ruby already seems to supports a couple of the modifications I suggest, though the documentation is sketchy.

Arcane parameters

# create a sniffer that grabs the first 1500 bytes of each packet 
$network = Pcaplet.new('-s 1500')

Bloody Perl people. Variables prefixed with $ are global in Ruby and a bad idea to use willy-nilly. But that’s a side point to this line: what the hell is that ”-s 1500” all about? Arcane command-line arguments have no place in a scripting language object class. Much better to use named parameters or a method call to set the grab length.

network = Pcaplet.new(grabsize => 1500) 
# or 
network = Pcaplet.new().grabsize(1500)

Where’s a delegating factory when you need one?

# create a filter that uses our query string and the sniffer we just made 
$filter = Pcap::Filter.new('tcp and dst port 80', $network.capture)
# add the new filter to the sniffer 
$network.add_filter($filter)

(More globals. Ick.) What’s $network.capture and why is it being passed to a Pcap::Filter constructor? Why do you have to add it to the $network object later?

Why not just use a delegating factory (if that’s the right term)?

network.add_filter("tcp and dst port 80")

Let the network object figure out that you’ve passed a string, create a filter object with whatever arcane parameters it needs, and attach it. Less thinking for me, happiness all round!

Update: Looking at the class code is amusing: if you pass a filter object to add_filter, it seems to grab the string you used to create it!

Ugly iteration

# iterate over every packet that goes through the sniffer
for p in $network

Ugh. Hidden iterators: bad. Single letter variables: bad.

for packet in network.packets # all packets 
# or 
for packet in network.captured # does captured imply filtered? 
# or  
for packet in network.filtered_packets # just the filtered ones, ta

Filtering confusion

# print packet data for each packet that matches the filter 
puts p.tcp_data if $filter =~ p

Why is it testing the packet data against the filter if it already attached the filter to the sniffer? Does adding a filter to the sniffer not actually filter the packets? (It does.) Later on, when there’s three filters applied, it makes sense as you don’t know why your packet has been captured unless you manually check against the filters but that just implies a callback-driven mechanism would be a good idea.

network.add_filter('tcp and src port 5190', proc { aim_recv })
network.add_filter('tcp and dst port 5190', proc { aim_send }) 
# etc.

More Constrained Tube Journeys

rjp — 2005-11-08T12:09:44+00:00

On the way back from New Eltham this evening, I started pondering some more constrained tube journeys:

shortest journey with at least one non-interchange (no circle on the tube map) station on each line (Waterloo & City exempted) starting and ending at an interchange (stations on two lines only count for one line at a time).
the longest journey only using interchange stations (black/white circle but only tube lines and with different lines than its neighbours—Farringdon doesn’t count, for example.)

For the first one, I can do it in 30 stations (solution later). For the second, the best I’ve got so far is 11.

[11:11, 8th November. 19 stations for the second one.]

dc mandelbrot generator

rjp — 2005-08-27T10:47:29+00:00

I wrote this about seven years ago, probably something to do with my 1994 version that I wrote in FALSE.

2k64sw32sh7sz[.]0:p[,]1:p[:]2:p[o]3:p[*]4:p[O]5:p[@]6:p[ ]lz:p5lw/si4
lh/sj0 2-dsxsu0 2-dsysv[0 2-dsxsu[luSalvSblzdsnSq[lqSn0Sq]Sm[ladd*SoSs
LoLslb2** SoSsLoLslbd*-lu+Salv+Sbclad*lbd*+4tLrln;pPlilu+ Sulu2>h]sh lhx[]pljlv+Svlv2>f]sflfx

Put that in a file and “dc < file”.

Random idea #1: Autopano

rjp — 2005-08-27T09:47:54+00:00

use autopano to line up the two halves of a red-green stereoscope
use autopano to identify pictures which are similar

Diagram of the 14 stop solution

rjp — 2005-08-27T10:48:30+00:00

Constrained Tube Journeys, #3

rjp — 2005-08-27T10:52:38+00:00

Standing at East Putney after New Year, I thought of two more constrained tube journeys to try and figure out.

The trip enclosing the biggest area without backtracking
The shortest trip involving each line just once

For the first one, something like: Rayners Lane, Acton Town, Victoria, Stockwell, London Bridge, Stratford, Liverpool Street, Baker Street, Rayners Lane. Unfortunately, tubes.d doesn’t have the necessary information for a script to calculate the area enclosed by journey yet.

For the second (ignoring the DLR for now), something like: Shoreditch (ELL) Whitechapel (HamCity) Aldgate East (District) Tower Hill (Circle) Aldgate (Metro) Liverpool Street (Central) Bank (WaterlooCity) Waterloo (Jubilee) Green Park (Victoria) Oxford Circus (Bakerloo) Piccadilly Circus (Piccadilly) Leicester Square (Northern) Tottenham Court Road. A total of 14 stops with only 1 non-interchange. Since you need 13 stops as a minimum for 12 lines, that’s pretty good.

update, 20050102: A quick hack with Prolog seems to imply that there’s no 13 stop solution at all. More about this later.

Constrained Tube Journeys, #4

rjp — 2005-08-27T10:51:02+00:00

Is it possible to visit all the lines in alphabetic order? (i.e. Bakerloo, Central, Circle, District, East London, Hammersmith&City, Jubilee, Metropolitan, Northern, Piccadilly, Victoria, Waterloo&City)

Since there’s no interchange between the Victoria line and the Waterloo&City line, the answer is no. If you cheat and use the Waterloo&City line first (from Bank to Waterloo), the answer is, well, I don’t know yet. Maybe.

update, 0147: There’s also no sensible way of going District, East London, Hammersmith&City.

update, 0149: Although including the DLR would fix that problem and gives a possible solution (which this margin is too small to contain!)

Constrained Tube Journeys, #2

rjp — 2005-08-27T10:53:51+00:00

After finding sequences of alphabetically ordered tube stations, I wondered if there were any strictly alphabetical sequences. Modifying tubesequence.pl was fairly easy (the commented out code) but there’s only one: Ravenscourt Park, Stamford Brook, Turnham Green.

More Constrained Tube Journeys

rjp — 2005-08-27T11:02:01+00:00

Whilst standing at Piccadilly Circus waiting for a westbound Piccadilly Line tube, I noticed that the last six stops all began with the letter ‘H’ (Hounslow East, Hounslow Central, Hounslow West, Hatton Cross, Heathrow Terminal 4, Heathrow Terminal 1,2,3). That got me wondering what the longest sequence of alphabetically ordered stations was. One Perl script later (using tubes.d from earlier) and the longest sequences, all of 6 stops, are:

BakerStreet → BondStreet → GreenPark → HydeParkCorner → Knightsbridge → SouthKensington
BaronsCourt → EarlsCourt → GloucesterRoad → HighStreetKensington → NottingHillGate → Queensway
Borough → Elephant_Castle → Kennington → Oval → Stockwell → Vauxhall

Three sequences of length 6, none on a single line.

Constrained Tube Map

rjp — 2005-08-27T11:07:48+00:00

Given the distances between tube stations, can a tool like neato lay out the tube stations in anything like the correct layout? After typing in all the stations and distances, processing the data in graphviz format, and running it through neato, the answer is almost, but not quite.

[550k PNG] [SVGZ]