I’m so proud of myself. Over the years I’ve accumulated boxes full of old computer junk that I’ve been hanging onto on the off chance that I might need it sometime and using the rational “it cost my $$$$ five years ago it still must be worth something”. Well on the weekend I dragged out my boxes full of junk and started sorting through them.

I found:

• dozens of power cables, serial cables, printer cables, network cables, etc
• 2 x 14.4 Fax Modems (ext), 2 x 56.6 Fax Modems (int)
• About 8 old harddrives ranging from 125M – 8G (mostly on the lower side)
• several cdrom drivers, a 2x burner and a few floppy drives
• A case of old, mostly wrecked ZIP disks, Ditto tape drive with 2 tapes
• Original disks for stuff like PCTools for Windows, Turbo Pascal 7.0, etc.
• Enough X10 equipment to control every appliance twice in our Condo
• Old HP DJ340 Printer I used with my laptop
• Many old 3C509s a few Intel NICs and a 8-bit SNC NIC and 5 or 6 PCI NICs
• Old ISA Trident Video card, AGP ATI Card
• Several ISA sounds cards (SB 8-bit!), 2PCI Cards
• 2 ATX powersupplies, 1 P100 with Mainboard / 192M ram
• 3 wireless video cameras, 2 cordless phones
• Several switches and hubs, three joy sticks, 2 keyboards, four mice, etc, etc, etc.

From all my only conclusion is that I hang on to stuff too long. I picked through and pulled out some of the better stuff to keep (including my X10 stuff). The HPDJ340 seems like it should still be worth something so I’m going to give a shot to selling it on E-Bay. And fot the rest of this stuff I stuffed it into a few boxes and someone at CLUG offered to take it all off my hands. Yeah.

Since I have these old HDDs and I have no idea what’s on them I don’t want to give them out as is. You never know what kind of incriminating evidence is on these things like my old ENEL labs that proove I really didn’t understand my power courses… I found a bootable linux CD by Jetico (here) that automagically and securely wipes all the HDDs attached to the computer. This made it really easy to destroy all the drives before I give them out.

So now Anji and I have a whole tonne of extra space in our Condo storage closet for me to accumulate new junk

I was talking with Anji’s brother yesterday about his fourth year engineering project (he’s in Geomatics engineering). He was wondering about the difficulty of building a program that reads location information from a GPS receiver, takes any scanned / acquired map / lets the user pick two or more known points on it, and thereafter plots the current position on the map.

Not knowing anything about GPS or mapping I figure this shouldn’t be too difficult. I imagine reading from the GPS is simple. He was talking about a USB one but I think I’d stick with serial. The amount of example code involving serial port IO is pretty substantial compared to USB. So assuming that I can query the GPS device and pull things like the lat/long then we just need a GUI.

Anji pointed out that this GPS stuff is fairly difficult to do properly because of curvature of the earth and stuff like that. I figure though that unless you are trying to map your position in North America or something big like that we can probably get reasonably close by just making some simple assumptions. Namely:

1. Curvature of the earth is neglegable
2. Assume lat/long lines are straight in our small window

This means we can just to plain old linear interpolation to find our map position. ie)

def interp(x1,d1,x2,d2,x):
    return (x-x1)*(d2-d1)/(x2-x1) + d1

# training points
long1 = 10;     mapx1=100
lat1 = 10;      mapy1=10

long2 = 20;     mapx2=140
lat2 = 40;      mapy2=80

# current position
cur_long = 15
cur_lat = 50

cur_x = interp(long1, mapx1, long2, mapx2)     # = 120
cur_y = interp(lat1, mapy1, lat2, mapy2)       # = 103.333333333

print "Long %s, Lat %s maps to %s,%s on the map" % (
    cur_long, cur_lat,
    cur_x, cur_y)

So then all the work is just a simple matter of plotting a map on the screen and plotting these points on them. Since we know their pixel locations on the map this is easy too. I figure that, assuming the GPS data format is well documented, a first draft at this in Python would be about an hour or two work.

So the program logic could be something is simple as:

1. Startup: Let user pick setup points either by entering them or by going to that location and clicking on the map.
2. Main loop:
   1. Read lat/long off GPS device
   2. Convert lat/long to map x,y using the above interpolation calculations
   3. Draw map
   4. Draw position on map using calculated x,y
   5. Draw extra informational stuff like speeds, bearings, height, etc on map
   6. Repeat

This leaves some room for improvement like:

1. Better lat/long to map calculations that do account for curvature
2. Use more training points for better accuracy – easiest way probably just compute map x,y with each combination of training points and take an average (ditching the extremes).
3. Zooming and panning of map
4. Waypoints, etc

Anyways. Makes sense to me. I think this would be a reasonably accurate first pass at it. I’m rather tempted to try this. I wonder if I can purchase a cheap serial GPS receiver on E-Bay or similar.

Here are a few related links:

• A python library for converting LatLong’s from GPS device to UTM – This is pretty well documented code for accurately converting LatLong coords to x,y coordinates onto a map
• PyGPS – Is a similar program written by the man, Russell Nelson, for an iPAQ type device. It seems a little more correct than my simple model here
• Commercial serial port programming library for Windows – A well documented interface to the serial port under windows (for talking to GPS devices). It’s free for personal use.
• A slick Serial library for Python (Win + Lin) – Here is a nice userfriendly looking serial library for Windows and Linux.
• GpsDrive – Is a very snazzy looking GPS Navigation program written with GTK on Linux. Supports a variety of GPS devices and neat nav features as well as autofetching maps from the Internet.
• Another very capable looking Windows Serial IO library

Kyle pointed out a new spam prevention technique called grey listing. Basically it operates off the premise that 95% of spam sent is just fired off and ignored (ie. doesn’tretry on errors, stuff like that).

Basically what grey listing does is for each inbound message it checks to see if the combination of sender, recipient and senders IP have ever sent a message through this server before. If it hasn’t then the message is rejected with a temprorary failure for about on hour. Normally SMTP servers will retry periodically so this just means the first time you send a message to someone running a grey list it’ll take about an hour to deliver. After that it should be pretty snappy. Sounds like the author of this has some pretty promising numbers suggesting that this really works.

I see a few problems:

1. One hour latency on messages might be a bit steep
2. Some mailing lists and users have dynamically generated e-mail addresses. For example EZMLM. This would mean each message is going to take an hour to deliver!
3. Problem with backup MX. I control my primary server but not the backup MX server. This means I can install greylist software on my server but not the backup MX. So what would happen is spammer sends message to primary which produces a failure so spammer tries backup MX, which accepts the message and promptly tries to send it to my primary which temporarily rejects the message but since the backup MX knows how to retry it’ll keep trying until the primary accepts the message and the spam gets in anyways. So unless you operate both servers I don’t think this really would work. Too bad. I think 1 and 2 I could live with but not this.

This methods sounds very promising because it requires zero client setup and almost no work by the admin.

Last night was kind of cool. Anji works Monday and Wednesday nights at SAIT so it was just Kailey and I last night after dropping off Anji. We ran to Future Shop to complain about some digital prints we had made and then headed home. She fell asleep in the car on the way home so when I pulled into the parkade I decided this was a great time to clean out the car a bit (our car has been accumulating stuff like jackets, etc over the past few months). She woke up just as I finished so we headed up to our suite.

Kailey set in her car seat for a bit longer and watched me do some dishes and put away the stuff I dragged up from the car. Then we played for a bit and then she started screaming.

I tried rocking her, holdering her different, etc etc but nothing was working. I brought her out on the balcony and saw our neighbours going for a walk. I tried to talk to them but I couldn’t hear anything they were saying. To try and calm Kailey down I grabbed the Snuggly from the car and put her in it and we went for a short walk. It worked wonderfully. When we got back to the Condo she was still happy so I grabbed my book and walked around while reading with Kailey still in the Snuggly. She started getting fussy just as it was time to leave and pickup Anji so we went back into the car. She cried for a little bit but quickly fell asleep.

What a nice evening. Unfortunately winter is coming and walks are going to be more difficult to do.

So I found this article about a Quebec “woodsman” decided he wanted a bear as a pet. Here is an excerpt from the article.

The cub, nicknamed Buddy Bear, was swimming across the river near Wakefield, north of Ottawa, when Denis Ryan grabbed it.

The bear broke free several times by clawing at Ryan and tried to swim to shore, only to be dragged back out into deeper water.

To wear out the bear, Ryan ran over it with the Jet Ski, forcing the cub’s head under water.

The 55-year-old woodsman got his best grip on the cub by holding it upside down by one of its hind legs. He then dunked the animal repeatedly to drain the cub’s energy.

The cub was moaning, desperately trying to breathe.

“I just lifted him up and then I could dunk him. Then he couldn’t breathe. I kept dunking and kept dunking him,” Ryan recalled.

“I was never mean to the bear. There was a couple of times I wanted to hit him over the head with a pipe or something but I didn’t do that.”

To stop the cub from jumping off the Jet Ski, Ryan tied a rope around one of its hind legs, kept one hand on a leash, the other on the handle bar.

This guy is some sort of messed up redneck freak. Bears are very large, very dangerous animals and they need their space. Trying to bring them into civilization as a “pet” is a stupid idea. The methods he used to tire out the poor cub were unspeakably cruel. If only someone would return the favour for this moron.

The really sad thing is that this freakshow gets off scott free because our laws define animal abuse only for owned animals like live stock and pets. Since the bear isn’t owned this episode is not animal abuse.

It’s nice that at least he was humane enough not to beat it with a pipe (insert large amounts of sarcasm here).

I came across this on the qmail mailing list and was rather amused

 $ ((RANDOM%6)) || rm -rf ~

A few weeks ago two teens took a 22 and started shooting at cars on a highway killing one person and injuring another. The families of the victims are being sued for $60M because these kids named their reason as “an attempt to recreate the images of Grand Theft Auto” a violent computer game.

This is retarded. One more example of the unwillingness of anyone in the US to accept responsibility for anything. It’s always the other guys fault. Yes, maybe playing violent video games 24×7 might cause children to become aggressive but seriously is it the video game manufactures fault? NO!

The real problem here is bad parenting. When I was younger Wolfenstien 3D was becoming popular. I loved that game and played it often but always in moderation. My parents realized that letting me sit in from of the computer, playing video games, violent or not, was a bad idea. Sometimes when I played for hours straight I would get really angry at the slightest distraction like chores or whatnot but do you think it was the video game? NO! The same thing happens if I sit in front of the computer coding for several hours. It just isn’t good for someone to sit there day in day out. I’ve since played many other violent games but yet my real life frag count is still zero.

Bad parenting is to blame for the fact that these children figured it was a good idea to snipe cars with a 22. I wouldn’t be at all surprised if this had nothing to do with GTA and they just blamed it on the video game because everyone else is doing it.

I bet if one looks at the what these children did with their time you would probably find that they didn’t just play the game a bit, they probably played a lot with no supervision. Video games are great but parents have to make sure their children get a healthy dose of reality as well.

Anyways I’m just getting sick of all the stupid lawsuits the US is letting slide. People should be accept responsibility for what the do not blame the company because they screwed up or were careless.

So with all the stupid things happening lately like the SCO lawsuits, the patent debates in Europe, the RIAA, etc it seems Verisign decided to jump on the bandwagon and be an idiot too. (here is a good review of this mess.)

Verisign runs the root name servers for the common TLDs like .COM, .NET, etc. In the past if you try to lookup a name and it doesn’t exist the DNS servers would return an error saying that which would show up in your mail bounce message or web browser or whatever. Makes sense. Incidentally this behavior is what makes many Spam filters work as often the spammers will generate fake domains in their from address and many anti-spam tools would bounce messages that contain addresses that don’t resolve.

Anyways now Verisign decided they could make more money by actually having unknown domains resolve to one of their servers where they can sell advertising space on the website and, even worse, accept SMTP messages to harvest the e-mail recipients addresses. This is going to confuse a lot of people really screw up a lot of software now that a typo will now resolve to the Verisign server.

To make them seem even stupider they decided that, instead of using a real SMTP server and telling it to reject connections, they should use some program called “Snubby Mail Rejector” that returns a fixed set of responses and ignores the input (ie. it doesn’t even attempt to follow the SMTP spec – here is a rant about how stupid this is). ie)

$ telnet dlkfjdflkjdf 25
220 snubby4-wceast Snubby Mail Rejector Daemon v1.3 ready
HELO myhost.com
250 OK
MAIL FROM:
250 OK
RCPT TO:
550 User domain does not exist.
DATA
250 OK
...
221 snubby4-wceast Snubby Mail Rejector Daemon v1.3 closing transmission channel

This almost looks reasonable. But wait. What if we try typing in junk instead?

$ telnet dlkfjdflkjdf 25
220 snubby4-wceast Snubby Mail Rejector Daemon v1.3 ready
asldjsljslkdjklj
250 OK
ksaldkjsdlkj
250 OK
sdsdlksjdlksjdk
550 User domain does not exist.
skjldlskjdlksjdlkj
250 OK
kljsdlkjslkjd
221 snubby4-wceast Snubby Mail Rejector Daemon v1.3 closing transmission channel

At the moment they are rejected the bulk of the message but what if they change it so that the bounce it after receiving the whole message? This would mean that they would have access to enormous quantities of e-mail which they could process for addresses or worse. It seems the Verisign has already thought of this and is planning on collecting all e-mail’s (with body) sent to their machine.

On of the really bad things is what happens with typos in the primary MX record. Generally a domain will have a primary MX record and then a series of secondary MX records which are the machines that accept e-mail in the event the primary is down. In the past if I was editing my MX record for the primary and made a typo all the SMTP servers trying to send me e-mail would realize the primary is down and use the secondary and so on. It just works. Unfortunately now if you make a mistake in your primary MX record it will resolve to a machine that does run a mail server that will reject all mail meaning that it will NEVER get to your backup mail servers. This is really stupid!

On an amusing note it seems that Verisign may have underestimated the number of typos being made as their server is just barely responding to web requests.

In summary. The Internet was way cooler before companies got involved!

Luckily for us many of the name server packages are already patching their software to undo this mess and ignore certain IPs like this one. Unfortunately there is no good way of really fixing this.

• Patch for DJBDNS (TinyDNS)
• Patch for BIND

Here are some other relevant links:

• here is a petition you can sign to protest this gross misuse of the DNS system.
• Netster is suing Verisign for antitrust violations and unfair competition. Finally a useful lawsuit. (here is the companies official page on this)

Version 1.5 of jmlm.py fixes a few problems that were preventing jmlm.py from running on certain POP servers as well as a bug where the SMTP server does not add a “Return-path” header.