lerfjhax.com

Mortgage Shopping in Chicago

Mon, 28 Jan 2008 19:18:00 EST

I’m in the process of buying a home in Chicago, and as all informed buyers should do, I went to several different lenders to compare their offerings. Money is, after all, a commodity—everyone that sells mortgages does so with someone else’s money, either directly or indirectly, so they’re no competition on product, only the related service.

For those keeping score: I have excellent credit, am putting 10% down, and want a 30-year fixed mortgage. My income can easily support those payments, so there’s absolutely no problem getting financing. The only question is how good my financing package will be.

Less than 20% down in the US means you need either private mortgage insurance (PMI) on a single loan (90%), or two loans (80% plus 10% at a higher interest rate), since lenders won’t front 90% without PMI.

I filled out a form on a website that puts me in touch with a bunch of different lenders, and followed up on referrals from friends and family. Ultimately, there were five contenders.

I gave them a feel for my situation late December and got initial estimates from each. I was very clear with each of them that I was shopping around, so I would expect that they would give me their best offers if they were interested in my business. A couple weeks later, I went house shopping, and got a purchase agreement accepted. I called everyone back, gave them the details, and waited to see what they came up with.

My experiences follow. There isn’t a reproducible methodology here, so your mileage may—and very likely will—vary.

WCS Lending

In my opinion, there’s nothing to really differentiate them from the pack. The rates they were offering were competitive with (though not better than) anyone else, and while I developed a working relationship with my broker there, there was simply no reason to choose them over the others.

I asked for pre-approval letters of vary amounts, which were delivered promptly. My contact there spoke with me whenever I tried to call him, even over the December holidays and after-hours. He was eager to help, but seemed removed from the actual workings of the mortgage market, offering few suggestions and frequently needing to look into a question and call me back.

Chicago BanCorp

The broker I worked with seems to confuse his job with selling used cars. He himself called me a “a lot more savvy than a typical first-time home buyer”, but he didn’t seem to take that to heart.

The initial estimate included four scenarios: with and without PMI, and each of those with and without bank fees. The bank fees totaled $1000, which are somewhat significant, but not that significant considering I’m putting 10% down on a large mortgage anyway. He’d eat the bank fees for me—in exchange for bumping the rate a quarter point.

Does he think I’m stupid? If I’m putting down tens of thousands of dollars already, would I choose to save an extra thousand dollars now in exchange for tens of thousands in more interest? Insulting.

He sent me the document, and asked what I thought. I responded simply that I had been offered a rate .375% lower earlier that day. His response? “I have this conversation ten times a day. If I saw a gas station at $3.00/gallon, and one at $3.20/gallon, and one at $1.50/gallon, something’s wrong with this picture. I certainly wouldn’t put that in my car!” What I didn’t tell him is that everyone gave me a rate .375% lower.

I gave him the courtesy of competing with the others once I had a final address and price. He came back with “I can’t believe how much the market has improved!”, and still managed to be a half-point higher than the others at best (the PMI + bank fees option). I pointed this out, and he offered to reduce by a quarter-point “just for me”, but no more, because then he’d be underwater.

Seriously: used car salesman.

Stay tuned for part 2.

Space Shuttle Viewing Process

Thu, 6 Dec 2007 18:30:00 EST

Wait until space shuttle launch transportation tickets go on sale.
Buy tickets.
Reserve some photography equipment.
Wait several weeks.
Wake up at 5 AM.
Meet four friends 15 miles away at 6 AM.
Drive.
Drive.
Drive.
Arrive at Kennedy Space Center.
Get through security, explaining that the bag really is just one big lens, and that I really do need another bag to carry the rest of my gear.
Tour the museum of early space flight history, explaining everything in expert (excruciating?) detail for the rest of my group.
Get in line for an IMAX movie.
Discover that STS-122 has been postponed due to problems with the engine cutoff sensors. Déjà vu.
Depart Kennedy Space Center.
Drive.
Drive.
Drive.

That’s about where we are now.

Using Zebra Printers under OS X

Thu, 20 Sep 2007 17:09:00 EDT

You would think that a major printer manufacturer would bother making drivers for the second most prevalent desktop operating system. If you did, you’d be wrong. Zebra printers don’t have any OS X support at all—but that’s no reason to despair.

I find myself in a situation where I want to print to a Zebra printer under Mac OS X—specifically, an LP2844—for high-volume printing of FedEx shipping labels. Don’t tell me “businesses run on Windows™”, “just buy a Dell!”, or any of that. Yes, it’s true that most shipping software runs exclusively on Windows. (It’s also true that most shipping software sucks.) The point is, the business in question already runs on Macs, and I’m not about to change that.

Sure, I’m swimming upstream, but I’m just that kind of guy.

Here’s what I knew before I got a printer to play with:

FedEx has a SOAP API. Among other things, this lets me create shipments and get labels returned as PDFs or ZPL (the native language of Zebra printers) regardless of platform.
There are rumors on the CUPS mailing list of people that have gotten Zebra printers to work under OS X.
Endicia for Mac (USPS shipping software) supports the LP 2844 over USB, which means there’s some way to talk to this printer using USB from user space.
NRG Software makes an application that can monitor a directory and send files to a label printer, but I’d rather not pay $150/seat unless I have to.
There are plenty of print servers that talk LPD on one end and parallel on the other, so worst-case, I can throw ZPL at this printer over the network, even without OS X’s cooperation.

Ideally, I’d be able to talk to this printer over USB in such a way that Endicia still works too. After all, you never know when you’ll want to USPS something.

Upon receipt of a printer and a roll of labels, I plugged it in, and found that Mac OS X sees a printer. One hurdle down. Print using… wait, there’s no Zebra option. Well, that figures, since Zebra doesn’t make OS X drivers.

Ah, look. Easy Software Products (ESP) has a Zebra ZPL Label Printer driver. Neat. Printer added.

Okay, try to print… nothing. Not from OS X, not from lpr, not even lpr -l, which skips the printer-specific driver and passes the bits straight to the printer. Dang.

Anyway, I try some other things, including hooking it up to a print server, and get nothing. “How is this possible?”, I think to myself. Then it hits me: wait a minute, the LP 2844 doesn’t support ZPL. It talks EPL.

I get an EPL file to try, zing!, prints out great. Plug it over USB, zing!, works fine. After some fiddling with /usr/libexec/cups/backend/usb, I’ve got an XML-RPC service that runs on my Mac, returns a list of directly-attached USB printers (including serial numbers), and can send preformatted data (e.g. EPL or ZPL) to any of them over the network. It plays nice with Endicia, plays nice with multiple printers, requires no client-side configuration—you don’t even need to add the printer—and didn’t cost me a cent.

Good times.

Technical Bits

Here’s the gist of how I’m using the CUPS USB backend:

$ /usr/libexec/cups/backend/usb 
direct usb "Unknown" "USB Printer (usb)" 
direct usb://Zebra%20/LP2844%20?serial=_____ "Zebra  LP2844 " "LP2844 "

That enumerates attached printers, and gives you a URL that the USB backend can use to find the printer again. To send a print job, you say:

$ DEVICE_URI="usb://Zebra%20/LP2844%20?serial=_____" /usr/libexec/cups/backend/usb 0 nil nil 1 -

It takes EPL (or whatever) on standard input. I don’t really know what the arguments are for; a cursory examination of the source indicated they weren’t interesting to me for my application, it might be different for yours. Also, I looked only at usb-darwin.c, which is assuredly different than the usb.c for other platforms.

Given this, it’s a hop, skip, and a jump to making this network-capable, scripted, or whatever. If anyone is interested, I’d be happy to sell you my code (or build something special); email me at zebraprinting@<this domain>.com.

OCFS2: Racy rename(2)

Mon, 21 May 2007 15:52:00 EDT

On UNIX systems, the rename() system call is atomic. That is, once written to a filesystem, a file can have its name changed within that filesystem (including from one directory to another) as a single uninterruptible operation that will either succeed or fail but not result in a half-complete operation.

OCFS2 is a cluster filesystem that is now part of the mainline Linux kernel. It’s used in certain environments where multiple computers are plugged into the same storage device, typically a SAN. It was originally developed by Oracle, but is small in terms of kernel modifications and general enough in scope to be used by other people, so the kernel folk (including Linus) merged it into the tree a while ago.

And guess what? It’s rename() operation has a race condition. It’s still atomic, mind you, but it suffers from random failures.

The Race Condition

If two processes on different nodes of the cluster attempt to rename different files to the same filename, one of them may fail with EACCES.

This manifests itself as spontaneous permissions errors.

The Cause


/usr/src/linux-2.6.21.1/fs/ocfs2/namei.c +1213
        /* In case we need to overwrite an existing file, we blow it
         * away first */
        if (new_de) {
                /* VFS didn't think there existed an inode here, but
                 * someone else in the cluster must have raced our
                 * rename to create one. Today we error cleanly, in
                 * the future we should consider calling iget to build
                 * a new struct inode for this entry. */
                if (!new_inode) {
                        status = -EACCES;

                        mlog(0, "We found an inode for name %.*s but VFS " 
                             "didn't give us one.\n", new_dentry->d_name.len,
                             new_dentry->d_name.name);
                        goto bail;
                }

I’m not sure how this can happen, but I’m pretty sure this is the cause. OCFS2 has a master renaming lock, and it seems there’s an issue where the rename operation can change shape between figuring out what to do and acquiring the lock.

The Fix

I have no idea. Ask the OCFS2 guys.

The Workaround

Retry the rename() call if it returns EACCES up to some limit. There’s no real way to distinguish between a racing rename() and a genuine permissions error, other than the rename() will eventually succeed if retried.

And yes, this was fun to figure out.

Afterburners and Shock Diamonds

Wed, 9 May 2007 22:57:00 EDT

I didn’t see this photo when I went through my air show collection the first time, but I think it’s safe to say that it has something to do with reviewing 800 some photos before this one. Whatever the reason, I’m glad I found it.

The folks on Flickr seem to like it too.

Big Air Show Today

Sat, 5 May 2007 20:53:00 EDT

Good times were had by all. Except for the 90+ heat index, off-the-scale UV exposure, and ridiculous overcrowding at the beach. While my memory of the conditions will fade, my photos will not.

Check out my Flickr set, or watch the slideshow.

Big Air Show Tomorrow

Fri, 4 May 2007 21:13:00 EDT

My Canon 300mm f/4L IS and 1.4x teleconverter arrived today from LensRentals.com, no thanks to USPS, in preparation for said air show. LensRentals.com shipped promptly and with adequate lead time, but my postal service sucks. Lost mail, misdelivered mail (someone in another building got my car registration!), and the worst parcel handling ever.

Anticipating their usual tactics, I placed a sign on my door that read “USPS: Knock on the door, and we will sign for this package.” I also got everything I needed for my wife to be home all day; she was sitting on the couch near the door crocheting an afghan from 9 to 5. Despite this, when I checked the mailbox around 3 PM, I received a “Sorry We Missed You!” note. USPS simply does not bother attempting delivery of packages, and today was no exception. On the plus side, the wait at the post office at 5 PM was only 25 minutes, down from their usual.

In other news, F-15s circled overhead for at least half an hour tonight, no doubt due in large part to my proximity to Ft Lauderdale International.

I took pictures until the light wouldn’t let me shoot jets.

I’m looking forward to the air show. It’s going to suck—upper eighties, humid, odds of scattered thunderstorms approaching 100%, and abysmal parking situation—but I expect it’ll be worth it.

DotA RSS

Mon, 19 Mar 2007 07:20:00 EDT

For the uninitiated, DotA is “Defense of the Ancients”, a Warcraft 3 map that changes enough about the game to be completely separate game in its own right. It’s awesome. Check it out at their official homepage, and also check out the video below.

For those who have been playing it for a while, you will have undoubtedly found yourself in a situation where you log onto Battle.net and find a bunch of games using a new map that wasn’t out yesterday. Join, and you’ll hear a chorus of “kick DLer!!!11lolz” from the people already in the game. You may even find hosts who delight in letting you download 97% of the map only to kick you before completion.

No longer. I’ve long had a set of scripts to grab the latest DotA from getdota.com and automatically download it to all my computers. I’ve reworked these scripts to also publish a DotA RSS feed at:

http://itwasnt.us/dota/rss/

You can take this RSS feed and put it into your local RSS reader (I use NewsFire), or sign up for a service like Google Reader. Yahoo! even has an Alerts service that will send text messages to your cell phone for free, in addition to more conventional things like email.

Enjoy!

Florida is Weird

Wed, 21 Feb 2007 12:31:00 EST

A fight between a married couple ended in a death in Southwest Miami-Dade.

Sky 10 was above a home on fire in the area of Southwest 142nd Avenue and 38th Street. Police said they believed a woman set herself on fire in a fit of rage in the couple’s garage after telling her husband during a marital spat, “I’ll show you.”

She died at the scene, Local 10 learned.

Her husband has been transported to Jackson Memorial Hospital suffering from smoke inhalation.

It reads like a joke, but no, it’s actually a real news story. Florida is weird.

It's Comcastic! Still!

Mon, 19 Feb 2007 11:16:00 EST

Since I put in my monitoring system:

Down from 2007-01-31 02:42:20 to 2007-01-31 02:43:30
- 2007-01-31 02:42:40: Started Unicast Maintenance Ranging – No Response received – T3 time-out (code R005.0)
- 2007-01-31 02:43:13: Received Response to Broadcast Maintenance Request, But no Unicast Maintenance opportunities received – T4 timeout (code R004.0)
- 2007-01-31 02:43:17: No Ranging Response received – T3 time-out (code R002.0)
- 2007-01-31 02:43:22: DHCP WARNING – Non-critical field invalid in response. (code D003.0)
Down from 2007-02-06 07:03:29 to 2007-02-06 12:35:10
- 2007-02-06 12:33:59: SYNC Timing Synchronization failure – Failed to acquire FEC framing (code T002.0)
- 2007-02-06 12:34:13: SYNC Timing Synchronization failure – Failed to acquire QAM/QPSK symbol timing (code T001.0)
- 2007-02-06 12:34:47: No Ranging Response received – T3 time-out (code R002.0)
- 2007-02-06 12:34:52: DHCP WARNING – Non-critical field invalid in response. (code D003.0)
Down from 2007-02-12 07:03:30 to 2007-02-12 09:17:50
- 2007-02-12 09:17:22: SYNC Timing Synchronization failure – Failed to acquire QAM/QPSK symbol timing (code T001.0)
- 2007-02-12 09:17:22: SYNC Timing Synchronization failure – Failed to acquire FEC framing (code T002.0)
- 2007-02-12 09:17:23: SYNC Timing Synchronization failure – Failed to acquire QAM/QPSK symbol timing (code T001.0)
- 2007-02-12 09:17:26: No Ranging Response received – T3 time-out (code R002.0)
Down from 2007-02-16 07:03:20 to 2007-02-16 08:06:20
- 2007-02-16 08:05:50: SYNC Timing Synchronization failure – Failed to acquire FEC framing (code T002.0)
- 2007-02-16 08:05:56: SYNC Timing Synchronization failure – Failed to acquire QAM/QPSK symbol timing (code T001.0)
- 2007-02-16 08:05:57: SYNC Timing Synchronization failure – Failed to acquire FEC framing (code T002.0)
- 2007-02-16 08:05:59: SYNC Timing Synchronization failure – Failed to acquire QAM/QPSK symbol timing (code T001.0)
Down from 2007-02-17 07:03:30 to 2007-02-17 09:53:10
- 2007-02-17 09:52:35: SYNC Timing Synchronization failure – Failed to acquire QAM/QPSK symbol timing (code T001.0)
- 2007-02-17 09:52:36: SYNC Timing Synchronization failure – Failed to acquire FEC framing (code T002.0)
- 2007-02-17 09:52:37: SYNC Timing Synchronization failure – Failed to acquire QAM/QPSK symbol timing (code T001.0)
- 2007-02-17 09:52:50: No Ranging Response received – T3 time-out (code R002.0)
Down from 2007-02-19 07:03:30 to 2007-02-19 11:03:20
- 2007-02-19 11:02:44: SYNC Timing Synchronization failure – Failed to acquire QAM/QPSK symbol timing (code T001.0)
- 2007-02-19 11:02:44: SYNC Timing Synchronization failure – Failed to acquire FEC framing (code T002.0)
- 2007-02-19 11:02:45: SYNC Timing Synchronization failure – Failed to acquire QAM/QPSK symbol timing (code T001.0)
- 2007-02-19 11:03:01: No Ranging Response received – T3 time-out (code R002.0)

That’s four outages in a seven day period. All outages longer than 70 seconds started between 7:03:20 and 7:03:30 AM, and my monitoring system has a ten-second resolution.

I’ve called about every outage. I’ve gotten names, Comcast IDs, and/or phone numbers from everyone I talked to. I’ve asked for a supervisor, been told no one is available, been added to a queue for callback, and have received zero phone calls from Comcast. All they’ve done since December is send out a tech who took a printout of my records to take to his supervisor, who also promised a call I’ve never received.

I feel like this is the sort of thing that the people down at city hall would like to have on file for when it comes time to renew Comcast’s cable operator license.

Palm Trace Landings

Thu, 8 Feb 2007 09:28:00 EST

Allow me to share the experience with my apartment complex over the last 14 hours.

Yesterday, I got this wedged in my door:

They didn’t knock when they dropped it off, and I didn’t discover it until after the management office (and my bank) were closed. I did, however, confirm that my rent check cleared on February 5.

Early this morning, my bank sent me a copy of the front and back of the check, as well as some gory bank numbers detailing who and what took the money on the check. What do you know:

Endorsed by my landlord, and dated February 2 by their bank.

Just a few minutes ago, the management office opened. I was there, and so were four other people, notices in hand. I heard someone say “we sent out a hundred of those yesterday!”—gee, you think maybe if you’re sending out a hundred nonpayment notices, then you’re the ones doing something wrong?

It's Comcastic!

Tue, 30 Jan 2007 14:25:00 EST

Today makes ten outages in nine weeks. Not just “you’re throttling me” or “you’re resetting my TCP connections at random”, either—I can work with those. I’m talking about total service interruptions affecting me, my entire apartment complex (twenty-odd buildings, who knows how many subscribers), and probably the better part of the city.

Today’s was from 7:05 AM to 10:36 AM. It’s not a cable break… at least, not a customer-facing cable. Broken cables act like weak antennas, picking up enough RF to give me some kind of signal on broadcast channels. I don’t get any of that, which leads me to believe that the cable is being actively grounded. I don’t get signals at all, TV or Internet.

On top of that, Comcast service is terrible. They’re 2 for 5 on having a supervisor call me back. The customer service reps make up details to give to customers. They offer meaningless solutions: one supervisor offered to send out a tech or refer me to the network group so they can test my line. “I was told earlier that this is an outage affecting thousands of customers.” “Right.” “So, how will looking at my equipment or testing my line help?” “Looks like we can have someone out Saturday…”

Bell South won’t sell me DSL without a land line. I don’t want a land line; my wife and I use cell phones, and it works great. The person I talked to said they don’t offer metered service or emergency-only any more, so I’ll have to buy a full local line ($20/month plus who knows how much in taxes) just to get DSL on top of it. For that price, it’s cheaper to grab the laptop, drive out to Panera Bread and buy a few meals a month to work while waiting for the cable to come back up.

Sucks.

UPDATE: Another outage, this time affecting only Internet access, from 4:48 to 4:58 EST.

UPDATE 2: I’ve gone ahead and made monitoring scripts that poll various things across my line and poll my cable modem itself. The resulting output looks like:

Up from 2007-01-30 19:12:30 to 2007-01-31 02:42:20
Down from 2007-01-31 02:42:20 to 2007-01-31 02:43:30
- 2007-01-31 02:42:40: Started Unicast Maintenance Ranging – No Response received – T3 time-out (code R005.0)
- 2007-01-31 02:43:13: Received Response to Broadcast Maintenance Request, But no Unicast Maintenance opportunities received – T4 timeout (code R004.0)
- 2007-01-31 02:43:17: No Ranging Response received – T3 time-out (code R002.0)
- 2007-01-31 02:43:22: DHCP WARNING – Non-critical field invalid in response. (code D003.0)
Up from 2007-01-31 02:43:30 to present

It’s pretty cool.

Silly Turtles

Fri, 19 Jan 2007 12:53:00 EST

They’ve done it again.

Not only is Og sitting precariously close to the edge of the rock, but Merv is crawling on top of him.

This wouldn’t be the first time Og has gone around balancing on things.

Ants

Sat, 13 Jan 2007 15:43:00 EST

I’ve got some ants. Not an awful lot, and they’re getting poisoned at this very moment, so it’s not anything to be concerned with. But, it is a good opportunity to break out my macro lens.

They’re very little ants, too.

Good times.

Kennedy Space Center

Mon, 8 Jan 2007 15:46:00 EST

I visited Kennedy Space Center on Saturday. Everyone, myself included, had their camera; I saw a half dozen people with a 5D and 24-70 f/2.8 or 24-105 f/4 and a fair assortment of crop bodies and lenses, but only a few Nikon shooters.

I have to say, you never really understand the size of a Saturn V until you’re standing underneath one.

Good times were had by all.

STS-116 Launch

Sat, 9 Dec 2006 21:08:00 EST

Space Shuttle Discovery launched today at 8:47 PM EST. And even though I live two hundred miles to the south, I caught a glimpse:

Sweet.

Liquid Water on Mars Right Now

Wed, 6 Dec 2006 13:10:00 EST

NASA TV is holding a conference right now saying that there is liquid water on Mars today. From 240,000 images taken by Mars Global Surveyor, areas of two different craters have had light-colored flow-like areas form in the years between images. The volume of water is estimated to be between 5 and 10 swimming pools in each flow.

Ken—the guy with the neck-beard—called this the “squirting gun” twice. Thought I would mention that.

UPDATE: More detail from the media event can be found at SpaceFlightNow.

F-15 Flyby

Sun, 3 Dec 2006 21:48:00 EST

I was riding shotgun down the Florida Turnpike (northbound near Mirimar) around 4:00 PM EDT today when I heard a noise out the left side of the car. It was loud, and I couldn’t identify it as a car or motorcycle or anything. I caught a glimpse of planes flying low and fast over the freeway, raised my camera from my lap, clicked on the power, and snapped a few pictures.

Four F-15s flying in formation, likely in preparation for a flyby of the nearby Dolphin Stadium. Later research indicates that there was a Dolphins vs. Jaguars game starting at 4:05 PM, which seems to confirm this suspicion.

My camera went from off to having captured six frames in a matter of two and a half seconds or less. I love this thing.

This was at extreme range, even for a 200mm lens on an APS-C camera sensor (like 320mm on a 35mm camera). Here’s a 1:1 crop:

UPDATE: Good catch, Troy. I saw two engines and two tails, assumed they were Navy planes, and wasn’t particularly awake when I posted this. Oh well

Light Box

Mon, 20 Nov 2006 13:05:00 EST

I put together a lightbox a little while ago. It’s crude, but produces spectacular results, especially considering it’s built out of trash and $4 worth of stuff from a local arts and crafts store.

So far, it’s mostly my wife’s turtle collection. But, anything else I might capture on white will be tagged as such in my Flickr account, and end up in this list.

New Lens

Thu, 16 Nov 2006 23:30:00 EST

I received a Canon EF 70-200 f/2.8 IS in the mail today.

I went and took a couple pictures from a third-floor balcony, including this one. The fire hydrant is about 120 feet away, and this is a crop:

I actually had to dial back the saturation to get the colors to look more reasonable. The greens were too vivid! Sharpness is amazing, too. Look at these plants, along with a 1:1 crop from part of the left one:

Not only does it double the value of my camera bag, it doubles the weight, too. This is a heavy lens. I got some filters, too—humorously, the 77mm polarizer cost more than my entire 50mm f/1.8 lens. Oh well. This thing is awesome.

Zune on CNN

Tue, 14 Nov 2006 13:00:00 EST

For those who haven’t heard, Microsoft is making an “iPod killer” called the Zune. Check out this interview on CNN. Do it now, I’ll wait.

Some of my favorite quotes from the video:

“It does a couple things that are a little bit cooler than the iPod, but a lot of things that probably aren’t.”
“And you can email with it, right?” No.
“Can you pull songs off the network, if you have it on a server somewhere?” No.
“If you have bought songs on iTunes, it doesn’t play here. Even worse, if you’ve bought songs on Napster or some of the former Microsoft-compliant devices, also doesn’t work here.”
“Who do they think is going to buy this?”
“Why don’t they get some decent design people to make things look better? It’s clunky!” “I, I… no comment.”

I particularly like how Microsoft’s idiotic PlaysForSure™ brand completely falls apart. If you buy Microsoft-licensed technology, under the PlaysForSure™ Windows Media™ umbrella, it won’t work when Microsoft makes their own player. I bet all their hardware and service partners are really happy about this.

Decide for yourself.

Update 11/15: Believe it or not, Zune is incompatible with Vista.

Tales from the Amazon

Fri, 10 Nov 2006 10:10:00 EST

Yesterday, I received three emails from Amazon.com in the space of an hour, asking me to review the same three items. I thought this was kind of weird, and got an Amazon.com customer service representative on the phone. I explained the situation, and was told to use the email form instead.

I fill out an email, stating:

I’ve received three emails from Amazon.com in the last 75 minutes, asking me to review the same products. Reference numbers 3701220, 3713330, and 3697740. The customer service representative I just spoke with suggested that these could be phishing attempts, but the headers clearly indicate that these originated from Amazon.com:

< snip >

How many more emails will I receive about these items?

I then received an email back:

Thank you for writing to us.

The e-mail you received was not from Amazon.com. We are investigating the situation, and we appreciate you letting us know that you received this.

For your protection, we suggest that you never respond to requests for personal information that may be contained in suspicious e-mail. It is best to assume any e-mail that asks for personal financial information (or web site linked to from such an e-mail) is not authentic.

If you did not click on the link in the fraudulent e-mail, your account at Amazon.com is fine—there’s nothing more you need to do. If you did click the link, but didn’t enter any personal information (such as your login or password), the phishers will not have your Amazon.com account information.

However, please know that if you ever respond to a phishing e-mail and do enter your Amazon.com login and password (or any other personal information) on the forged web site, the phishers will have collected that information and you should take appropriate action. We recommend that you update your Amazon.com password immediately, and, if you entered financial information, you may want to contact your bank or credit card provider.

If you encounter any other uses of the Amazon.com name that you think may be fraudulent, please do not hesitate to contact us again.

Thank you for contacting Amazon.com.

Right. My reply:

On Nov 10, 2006, at 9:05 AM, Amazon.com Customer Service wrote:
The e-mail you received was not from Amazon.com. We are investigating the situation, and we appreciate you letting us know that you received this.

Yes, it was. Look at the received headers:
Received: from mm-retail-out-1104.amazon.com (mm-retail-out-1104.amazon.com [207.171.165.136])
    by katana.lerfjhax.com (Postfix) with ESMTP id C55C28B0
    for <delta407@lerfjhax.com>; Thu,  9 Nov 2006 13:17:32 -0500 (EST)
Received: from retail-mail-app-5113.iad5.amazon.com ([10.217.42.40])
  by mm-retail-out-1104.amazon.com with ESMTP; 09 Nov 2006 10:17:34 -0800
Received: by retail-mail-app-5113.iad5.amazon.com
    id AAA-merchandizing-15914,1461; 9 Nov 2006 10:15:48 -0800

Received: from mm-retail-out-1104.amazon.com (mm-retail-out-1104.amazon.com [207.171.165.136])
    by katana.lerfjhax.com (Postfix) with ESMTP id 0BD10BE7
    for <delta407@lerfjhax.com>; Thu,  9 Nov 2006 13:33:27 -0500 (EST)
Received: from retail-mail-app-5105.iad5.amazon.com ([10.217.41.63])
  by mm-retail-out-1104.amazon.com with ESMTP; 09 Nov 2006 10:33:31 -0800
Received: by retail-mail-app-5105.iad5.amazon.com
    id AAA-merchandizing-27598,1297; 9 Nov 2006 10:32:37 -0800

Received: from mm-retail-out-1103.amazon.com (mm-retail-out-1103.amazon.com [207.171.165.135])
    by katana.lerfjhax.com (Postfix) with ESMTP id 4BF71C52
    for <delta407@lerfjhax.com>; Thu,  9 Nov 2006 14:12:51 -0500 (EST)
Received: from retail-mail-app-5112.iad5.amazon.com ([10.217.139.62])
  by mm-retail-out-1103.amazon.com with ESMTP; 09 Nov 2006 11:12:51 -0800
Received: by retail-mail-app-5112.iad5.amazon.com
    id AAA-merchandizing-04866,44; 9 Nov 2006 11:11:37 -0800
Either these were sent by Amazon.com, or Amazon.com has lost control over its own servers. There is no third option.

According to my mail server, these messages were delivered to it from machines identifying themselves as mm-retail-out-*.amazon.com, from the IP addresses fall within AS 16509. AS16509 is registered with ARIN as “AMAZON-02”, to:
OrgName:    Amazon.com, Inc. 
OrgID:      AMAZON-4 
Address:    605 5th Ave S
City:       SEATTLE
StateProv:  WA
PostalCode: 98104
Country:    US
These messages originated with Amazon.com, period. Either they were sent knowingly, in which case sending three of them in succession is a minor glitch, or they were sent unknowingly, in which case it seems plausible that Amazon is mass-mailing millions of others.

I am forwarding my original complaint, along with this response, to abuse@amazon.com.

Humorously, I received two replies from abuse@ within one minute of each other—one by Vinod K. and one from Syed Ibrahim. Both replies were duplicates of the original.

No matter. I take the IP address of the Amazon.com mail servers in question (from the top Received: header), and punch them into the Internet Search tool on the left. It sends me to AS16509. Click “WHOIS”, and I’ve got a phone number for the Amazon.com Network Operations Center. I give it a ring, navigate the phone tree, and end up with a tech who doesn’t ever interact with outsiders.

“How did you get this number?” That made my day.

Anyway, he’s contacting the people that run those mail servers, and I’m sure they’ll get their stuff straightened out. Good times.

Like, "Ha Ha" Funny?

Thu, 9 Nov 2006 23:00:00 EST

You’ve no doubt seen NASA’s Shuttle Carrier Aircraft. They have two. As a refresher, they are modified 747s whose mission is to shuttle the shuttle from one coast to another. They get a very economical 125 feet per gallon—that’s about two gallons to move forward one plane length—and look something like this:

Anyway, see the struts that hold up the space shuttle on the rear? The starboard strut has a little sign on the front:

Who says government employees don’t have a sense of humor?

Normal is Overrated

Mon, 30 Oct 2006 09:30:00 EST

I was just talking to someone about a networking problem I was diagnosing. I said, and I quote:

It’s definitely NAT-related, since they’re getting a RST back from our address, and we’re definitely not sending one.

Immediately after I said that, I asked:

...is it normal to draw that conclusion without actually having to think?

I’m pretty sure that most people do not have instincts relating to network address translation or mysterious TCP failures. That makes me abnormal.

Oh well. Normal is overrated, anyway.

100mm Macro

Fri, 27 Oct 2006 13:07:00 EDT

My Canon EF 100mm f/2.8 Macro arrived today. Huzzah!

It is, of course, a macro lens.

But, it also gives me some telephoto reach beyond my 17-85 IS.

Steve Ballmer

Thu, 26 Oct 2006 10:55:00 EDT

I stumbled across this earlier today:

It’s nice to see that he hasn’t changed much.

Am I the only one that gets creeped out by watching those? I feel like I need to… take a shower? Is that right? He seems like he’s about ready to give himself a heart attack, or something.

See also The Register’s article Microsoft’s Ballmer: chair-tossing potty-mouth.

Comment Spam

Wed, 25 Oct 2006 21:17:00 EDT

Various articles, includings T-Shirts, Smoking is Good For You, and a few others have started getting comment spam. A lot of comment spam—T-Shirt was getting a few per hour.

Suck!

My solution? Comment forms that work only if the user-agent is interpreting JavaScript. In Rails, this looks like:

<%= form_remote_tag :url => {:action => "comment", :id => @article}, 
                    :position => :bottom, 
                    :update => {:success => 'comments'},
                    :html => {:action => '/must_have_javascript_enabled'} %>

This emits:

<form action="/must_have_javascript_enabled" class="commentform" 
onsubmit="new Ajax.Updater({success:'comments'}, '/articles/comment/46', {
  asynchronous:true,
  evalScripts:true,
  insertion:Insertion.Bottom,
  parameters:Form.serialize(this)
}); return false;">

A comment spammer will POST to /must_have_javascript_enabled, whereas an interactive user-agent won’t POST at all: onsubmit will trigger an AJAX request, which finally POSTs to /articles/comment/46.

We’ll see how effective this is in the coming days.

LerfLine Returns!

Mon, 23 Oct 2006 21:06:00 EDT

I was talking to one of my friends sometime over the weekend. He was sitting in Panera Bread, iBook out, chillin’ after (before?) work.

He asked a question with a long answer, and I decided it would be faster to talk on the phone than to type over IM, so I gave him a call. We talked for a little while, at which point he announced that Panera Bread was closing soon, and that he wanted to order a bagel.

He was going to put me on hold, but I objected; instead, he put me on speaker in his pocket, and told me in no uncertain terms to stay quiet. (I don’t know why he would think such a thing… ) He ordered his bagel in peace, as I captured and recorded the experience for posterity on a whim.

It then occurred to us: I had done something similar ages ago, where you call a phone number and have what you say recorded and posted to a website. This system was operable only for a short while, and I don’t really know what happened to it. Anyway, this seemed like a good enough idea to implement, so I did.

Half an hour later, LerfLine was born. I don’t know what will become of it, but there it is.

If you’d like to join the ranks of the elite experimenters, email me. If not, check out some of the random, weird crap that goes on in our lives.

Intersection

Sat, 21 Oct 2006 23:43:00 EDT

What’s worse: that’s not even a complicated intersection.

I hate Florida.

Canon Lens Information

Fri, 20 Oct 2006 17:20:00 EDT

I’m trying to pick out a telephoto zoom lens, and I’ve been irritated for a while about how no photo site that I’ve seen presents all the information I’m interested in at once. So, I decided to do something about it.

Check out the new Canon Lens Information section.

The biggest thing, in my mind, is how none of the photo sites have up-to-date street prices (if they list any prices at all), and how none of the stores have enough information to make a decision about a lens. The next biggest thing is that it’s hard to compare all of your options at once—for instance, someone looking at a 17-85 IS would probably like to hear about the 24-105 L IS, but most reviewers and stores put them in completely different areas. No more.

Pick a lens. Say, the 50mm f/1.8. It’s a cheap, fast prime, so if you’re looking at it, you are probably interested in other fast primes, or other comparably-priced ways to cover the 50mm focal length (such as the 28-90 for $30 more). Note how all the Similar Lenses are actually similar to the 50mm f/1.8.

Now, click on the 50mm f/2.5 macro. It’s a couple bucks more expensive, but it does macro, so maybe you’re interested in other macro lenses? Note how all the lenses are actually similar to the 50mm f/2.5 macro. And, since everyone is on a budget, they’re sorted by price. Amazing!

Note also that it suggests the 70-200 Ls to anyone looking at the 70-300 IS USM. My software came to that realization in a fraction of a second, while it took me several weeks to conclude the same.

Lerfjhax.com: bringing digital photography into the 21st century.

Og

Thu, 19 Oct 2006 12:02:00 EDT

He’s just sitting there. Seems to be comfortable too.

Og, you’re a silly turtle.

Targus Circular Polarizer

Wed, 18 Oct 2006 13:45:00 EDT

While ordering a backup CompactFlash card, I saw a 67mm circular polarizer on NewEgg—from Targus (?!)—and decided to give it a shot. It’s so cheap, there must be a catch: a quarter of the price of entry-level circular polarizers from reputable filter companies like Hoya. On the other hand, Targus is a reputable company… for cases and power supplies and things of that nature.

Whatever. It’s about $6 after factoring in the reduced shipping, and NewEgg has a 30-day return policy. Add to cart, checkout. It arrived today.

Humorously, the cardboard in the package says “Features: includes protective carrying case”, but white tape has been applied to the outside of the package to cover over that. And no, it does not include a carrying case.

I open it up, and hold it up to a mirror. I can see through it either way, which shouldn’t be possible with circular polarizers. Hmm…

Despite this, it does seem to behave like a circular polarizer, as my autofocus and metering aren’t broken, and since orientation of the filter affects the image. Compare:

Rotate polarizer

Optically, it doesn’t seem to degrade picture quality, and I don’t notice any reflections off of it. I’ll take it into some harsh conditions (like my trip to Windley Key state park) and see how it does.

And, by the way, it’s like this item doesn’t exist at all. It’s really unusual. According to Google, the only mention of a TG-67CPL is on NewEgg.com, on a page that no longer exists (but is still cached). Freaky.

Update 2006-08-27: Seems it fell out of the cache. Good thing I grabbed a copy of the product page.

Tamper-Proof Elections

Tue, 17 Oct 2006 21:55:00 EDT

Election fraud is a serious problem, and the current system in the US is a black box that can’t readily be audited. Ron Rivest—the “R” in RSA, the algorithm that typified asymmetric encryption for many years—is one of many security experts who has approached the problem.

The properties of an ideal election system are contradictory. Voters should be able to verify that their votes are counted correctly, but should be able to have their choices kept completely private—in fact, some would say that voters should not be able to prove they voted a certain way, even if they wanted to. So, can both of these goals realistically be achieved without sacrificing each other?

Turns out, yes. This is possible with a simple paper system: no complicated cryptography, no complicated machines, no complicated rules. It satisfies both of the required properties (privacy and verification) and is in every way more tamper-resistant than current elections. It’s called ThreeBallot, and it works like this.

As the name implies, in this system, voters cast three ballots instead of one. These ballots are to be optically read (like those standardized tests where you fill in the bubbles). Each ballot contains an identical set of spaces, one for each candidate in each race that’s up for election. Voters fill in the ballots such that the candidate they vote for receives two votes, while the candidates that they did not vote for receive one vote. The distribution of marks across the ballots is completely up to the voter.

Note that electronic “touch-screen” voting systems could also be employed, such that the voter chooses candidates once, and the machine prints a suitable set of paper ballots, without requiring the voter to think about the mechanics of the system.

The election constraints—at most one candidate is voted for in each race—as well as the system constraints—between one and two marks for each candidate when totaled across all three ballots—are verified by a simple machine, called the checker. (Cryptographers are creative folk.) If the ballots pass, they are stamped with random unique ID numbers, which the voter cannot see, and placed in a holding area. The voter chooses one of the three ballots, and a copy is made. The voter verifies the copy, accepts it, and thereby transfers the three ballots into the ballot box in such a way that the three ballots can never again be reassembled into a whole.

When the polls close, the ballots are counted by machine, much as they are now. The results are the same as with conventional voting, with the exception that the winner in a ThreeBallot system has double the votes as in a traditional system.

I find this fascinating. Note that the voter has a copy—a receipt—of one of the ballots. The voter can make the receipt look like anything, marking or leaving blank as many or as few spaces in any order, without affecting his ability to vote. This means that the receipt doesn’t give any indication at all of how a person voted.

Now, the best part: verification. When counted, all of the ballots are read along with the ID numbers and made publicly available by the voting authority. Given his receipt, a voter can look it up, and verify that the receipt matches what was counted. Since there is one receipt for every three ballots, verification is limited to one in three ballots; however, since an attacker does not know which ballots are receipts and which are not, very small amounts of tampering (modifying or removing votes) are detectable.

Verification could be handled by the individual voters, and for that matter, “watchdog” agencies could help. Voters could submit copies of receipts to e.g. the ACLU, who would have the ability to independently verify accuracy without compromising the privacy of the vote. Extensions of this are largely left as an exercise for the reader.

It gets even better. Turns out that verification does not necessarily rely on a voter verifying his own vote, only voters verifying some vote. The paper goes on to suggest a receipt exchange system, which defeats another class of attack without compromising the integrity of the system.

Check it out for yourself, and contact your local politicians. This is one way to secure a voting process that isn’t terribly far-fetched.

Geek Shirts

Sat, 14 Oct 2006 16:45:00 EDT

Of course, there’s ThinkGeek.com. But, I went and did something a little different.

The Internet is broken, Free the Mallocs!, and more at Syswear. Check it out.

Hooray Photography

Fri, 13 Oct 2006 20:10:00 EDT

My Digital Rebel XTi is awesome. Some of my favorites so far:

I still want to get a lens that does long telephoto. I’m looking at the EF 70-300 f/4-5.6 IS USM, but I’m considering others. In the meantime, my EF-S 17-85 IS USM is serving me well, and is definitely appropriate as a general walk-around lens.

One of my friends has a Tamron 28-300 that he likes a lot. It’s a nice lens, but it’s as slow as my 17-85, which I find is frequently scraping bottom for light and might be a problem if not for the image stabilization. On the other hand, if I was looking to put on one lens and never take it off, a 28-300 would certainly be a contender.

Good times.

Minnesota ISP

Fri, 13 Oct 2006 19:41:00 EDT

It’s no secret. My favorite ISP in the whole wide world is located in Minneapolis. They’re called Vector Internet Services, Incorporated, or VISI for short. Know why?

“We Suck Less” was their slogan some time ago. It’s since fallen out of favor and is no longer used, but it’s still true. This sounds negative, but really, it isn’t. All ISPs suck. Seriously. Let’s investigate.

Bandwidth is more or less a commodity. It’s worse with DSL or T-1 circuits, because you have your pick of company to stick on the other end. For this reason, no ISP can charge too much more than anyone else without losing all sorts of customers. As far as most customers are concerned, bits are bits, and it doesn’t matter who they’re coming from.

The barrier of entry is low. Really low. All sorts of moderately nice networking gear hits the auction block on regular cycles. Put a router, a closet, a fiber drop, and a twenty-something together and you’ve got an ISP.

ISPs depend on other organizations in order to deliver their product. Unless you’re Level3 or another Tier-1 provider, you’re buying bits from someone else in order to re-sell them to your customers. Unless you own all the land between you and your customers, you’re paying someone else to give you a cable. If there’s a problem with either of those, it’s your problem, even if there’s nothing you can do about it.

Why do I like VISI? They suck less. (Really!) They have a phone number that you can call if you have a problem. Someone will pick up, and they will be able to solve your problem. Once, I called, and did everything from re-flashing a Cisco DSL router to setting up ICP with their caching proxy server to diagnosing an IP-fragmentation-related performance issue without being transferred once.

They’re UNIX geeks at heart. This means their services work all the time, they can do more with fewer people, they can spend more time expanding and less time administering, and so on. As UNIX geeks, they run a busy USENET hub, a UNIX shell server, and a series of other cool things.

They’re open about everything. To this day, I’m subscribed to some of their mailing lists. These messages say things like:

VISI’s connection with UUNet failed at 14:54 CDT today. Traffic for UUNet destinations is being routed across Global Crossing’s backbone. VISI is working with UUNet to address the issue, but currently we have no ETR. Routing redundancy is operating normally, and VISI is not aware of any customer impact.

They send out emails documenting every scheduled server or router reboot, including who is affected, why it’s happening, and what’s being done to address the issue later. It rocks. Compare to Comcast, who drops my service two to four times per month with no explanation and no ETR, and you begin to understand why I appreciate VISI so much.

Their techs are sharp. In my years supporting VISI customers and being one myself, I’ve never once needed to escalate in order to solve a problem. Also, I’ll often call them even when I’m pretty sure it’s someone else’s fault—they’re consistently able to identify the problem in 60 seconds or less, and they’re frequently able to address the problem with engineers in the other organizations long before I would have even gotten to a first-level tech by calling them directly. (And yes, I’m talking specifically about Qwest.)

In short, if you’re looking for DSL service, T-1 service, colocation, or anything like that in Minneapolis or St. Paul, call VISI. You won’t be disappointed.

Digital Rebel XTi RAW Editing

Tue, 10 Oct 2006 20:56:00 EDT

The Digital Rebel XTi is under two months old. This explains why a lot of software—on OSX, most notably Aperture and iPhoto—can’t understand its RAW pictures. Aperture says “unsupported file format”, and iPhoto says something to the effect of “I have no idea what these files are, so do something with them yourself”, and refuses to drop them into your library or remove them from the camera.

UPDATE 2006-11-04: Apple has released an update that adds XTi RAW support to Aperture and iPhoto. Even so, keep reading.

Like other current Canon digital cameras, it shoots .CR2 files. However, they’re not at all the same as CR2s produced by any other Canon EOS camera—the sensor is unique to the XTi, and since the RAW file is essentially a full dump of what’s on the sensor, it makes sense that it should be different.

My current process is to use EOS Utility, provided with the camera, to pull the photos off the camera and drop them in my Pictures folder by date. (Note to other camera owners: use EOS Utility to put your name in the EXIF tags of all shot photos. It’s pretty cool.) I’ve configured it to do this automatically when the camera is plugged in.

After EOS utility finishes pulling in my .CR2s, it fires up Digital Photo Professional. DPP, also provided with the camera, is Canon’s image browser/RAW converter. It’s kinda quirky for a Mac program—Command-W closes the adjustment window instead of the current photo, it won’t let you quit unless you open the main window, etc.—but it works. The results aren’t bad, but picture to picture use give other irritations. Namely:

Changing noise reduction amounts is done in the Preferences window, not in the photo adjustments pallete. It’s annoying to go back there every time you change the ISO on the shot.
Cropping images is done using the Trimming Tool, but this doesn’t update the image window, meaning you have to adjust image parameters looking at the whole image, instead of just the section you’re interested in.
It’s not possible to rotate images at all.

I recently downloaded Adobe Lightroom beta 4. It’s free until February 28, at which point the beta expires. This fixes all of the above woes, and then some. It gives much more control over exactly how the picture is “developed”, meaning you can use it for one-stop picture processing, as opposed to frequently breaking out into another image editor.

These photo came out of Lightroom, and I have to say, there’s no way DPP could have even come close. If you’ve got an XTi, give it a try.

Lens Rental

Tue, 10 Oct 2006 12:57:00 EDT

Apparently, no one rents anything camera related in Ft. Lauderdale. Further, I was able to find only two places in Miami that rent lenses.

First is One Source Studios. They have a bunch of stuff, including a great selection of Canon EF lenses. Notably absent in my mind are the Canon EF 24-105 f/4 L and EF 100-400 f/4.5-5.6 L IS USM, but it’s a studio, so I can’t fault them for running only primes. Prices are a little steep compared to some of the mail order places, but they actually have all of this gear.

Second is Carosel Studios. Their website sucks, but they rent Canon EF lenses. They have some zooms (but not the 100-400 L), and their prices are better than One Source Studios by about 20%.

I’m thinking about renting a lens for some turtle shots, and to evaluate some telephoto zooms before purchasing (such as the 70-200 f/2.8 L). Merv is way too skittish, making it hard to get full-frame pictures of him even at 85mm. Og is more relaxed, but f/2.8 would be a lot better than f/5.6 for underwater shots. We’ll see.

UPDATE October 25, 2006: My new Canon lens information section includes pricing and availability from several mail order lens rental companies. Check it out.

Hard Boiled

Fri, 6 Oct 2006 15:15:00 EDT

This is a weird egg.

Thought I’d share.

Playstation 3

Thu, 5 Oct 2006 21:50:00 EDT

We’re about six weeks away from the launch of Sony’s next-generation console. Everything’s going great for them, except for one little thing.

Who is going to buy one?

For the price of a Playstation 3, you can buy an Xbox 360 and a Nintendo Wii. The Xbox 360 has been on the market for quite a while now, and has plenty of established games already. The Wii is innovative, inexpensive, and offers intriguing gameplay possibilities that no one else has. The Playstation 3 has…

...fewer exclusive titles than ever before? Oops.
...raised the entry cost to never-before-seen heights? Yikes.
...been delayed repeatedly due to its Blu-Ray drive, the sole purpose of which is to create another VHS/Betamax fiasco? That’ll go over well.

I have more disposable income than most people I know. I enjoy gaming—I might consider getting a 360 if I had more time (and am seriously considering a Wii anyway), but there’s no way I would pay $500 for the cheap model of a console. Plus controllers. Plus an HDMI cable. Plus $60 per game.

And you know what? There’s a lot of people that feel the same way. According to a recent poll on Slashdot, about two thirds of the viewership find Sony’s price tag “outrageous”. Slashdot is your nerdy early-adopter type, too—if anyone’s going to buy a new gizmo, it’ll be them.

Read up on the PS3, and come to your own conclusions. Personally, I’d love to see it fail, and take Sony down with it. Enough with the format wars.

Florida Softshell Hatchling

Mon, 2 Oct 2006 10:34:00 EDT

Something caught my eye crossing the sidewalk on the north side of our apartment, and the rest is history.

Word Search Approaches

Sun, 1 Oct 2006 22:02:00 EDT

Building off the success of the word scrambler, I’m working on a Lerfjhax.com word search creator. You know, where you pick out words from a jumble of letters.

Over the past few days, ideas had randomly struck me about how to make word searches more difficult than the ones I had seen in the past. I’m not sure why these came to me, but ideas do that sometimes. So, I figured, why not make a free web-based word search creator for others to enjoy?

It’s not quite ready for public use yet. However, I invite you to check out a preview—this is a sample of the “Diabolical” difficulty setting. Again, it needs some work (it probably doesn’t work very well in Internet Explorer), and the difficulty needs some improving, but it should give you an idea of what to expect.

Stay tuned for more.

Oracle 10g RAC

Fri, 29 Sep 2006 17:53:00 EDT

For those of you who don’t know what Oracle 10g RAC is, allow me to give you an overview.

Oracle is a relational database server. It implements more-or-less standard ANSI SQL, so the syntax is largely the same as what is used by other relational databases (such as IBM DB2 or Microsoft SQL Server). As an application developer, you’re responsible for organizing the “real world” into a relational model—for instance, taking a physical box full of widgets, and issuing commands to the database to add a row for each widget into the “widgets” table.

The next part, 10g, is a version number, referring to the tenth major release of Oracle since forever ago. The “g” is just because they didn’t like just saying “10”; there aren’t any other letters to worry about. Same goes for Oracle 9i.

The next part, RAC, stands for Real Application Cluster, and is the reason for my involvement with Oracle. That’s where you take more than one server, plug it into the same set of hard drives, and run one database across all of the servers. This is a departure from the traditional model where only one server can provide access to a database at once, and is really the only thing that can justify paying tens or hundreds of thousands of dollars for a database.

Allow me to explain that last statement. RAC, like any type of multi-master replication, is an exceedingly hard problem to solve, especially in a ridiculously high-concurrency environment like a database. Essentially, you have to coordinate all of these different servers so that the hundreds of simultaneous users don’t make the servers step on each other’s toes. Hence the difficulty.

Companies run solutions like RAC when they don’t have any other options. If you need to have a single relational database that’s always available and always writable, it’s the only thing you can do. RAC allows an entire database server to suffer catastrophic failure without affecting access to the data. If you always need to be running, period, RAC is the answer.

(Note that RAC can also be useful if you have a single database that is so large and complex that a single computer cannot keep up with the load. However, given the pace with which cores are being added to CPUs, this is less and less of a selling point.)

When considering an Oracle purchase, ask yourself: do you always need to be running, period? Is five minutes of downtime acceptable to your business? Would five minutes of downtime cost your business tens or hundreds of thousands of dollars?

If not, you don’t need Oracle. You need PostgreSQL.

Never heard of it? Some purchase-manager-friendly bullet points:

It’s free. It costs $0.
It’s free. The source code is yours to take and keep, meaning you’ll never, ever, under any circumstance be tied to any vendor.
It supports ANSI SQL standards better than Oracle, and better than virtually any other database server.
Its performance is comparable to Oracle under most types of workloads.
With some additions, the EnterpriseDB people have made it nearly 100% quirk-compatible with Oracle.
You can pay people to support your PostgreSQL installation, just like Oracle.
PostgreSQL was patterned after Oracle’s MVCC concurrency model, so it works the same way as Oracle when multiple connections are changing the same information.

The only thing PostgreSQL doesn’t do is RAC. (Here’s where the five minutes of downtime comes in.) You can have only one database server running the database at any given time. If it goes down, you can bring up another one right away, but it will be unavailable during the switch.

That’s not to say you lose any data. Both PostgreSQL and Oracle (even non-RAC) support standby databases, where whenever someone changes something, the changes are automatically sent to another database in real-time. You can have the database server start on fire, switch to your standby, and not lose anything. But, you can’t have more than one server modifying the database at once.

That’s it.

Is it worth it?

Payday Loans

Thu, 28 Sep 2006 20:46:00 EDT

Who uses them? Seriously. What is the demographic? When is it a good thing to get a payday loan?

On the one hand, I suppose it’s not completely illegitimate. Assume you owe $250 that you can’t afford, and that your next pay day is a full two weeks out. You could go to a payday loan place, get $250, and pay back $250 + 20% = $300 two weeks later. A $50 fee is pretty heavy, but is in the ballpark of late fees on many credit cards, and is probably favorable to late fees and penalties for failing to pay your power bill.

Now, consider that really bad credit cards charge at most 27% APR, compared to 20% for two weeks, or over 250% APR. A cash advance on a credit card is a better option. One might also consider using the overdraft protection that most banks offer, which is at most 16%.

One could argue that if you don’t have a credit card or a checking account with overdraft protection, payday loans and pawn shops are really your only institutional options. But what about personal options? How does it happen that all of your relations in this world put together can’t spot you some cash for a little bit?

I hear that I’m a rarity. When I buy things, I always pay with a credit card. (In fact, the only exception I can think of are parking garages that only take cash.) I do that because it’s more convenient than cash, faster than checks, and it’s safer for me than every other form of payment. But, get this. Every last one of my purchases is backed with cash, always. I have never once carried a balance.

You see, therefore, why I don’t understand payday loans.

When Educators Go Bad

Wed, 27 Sep 2006 22:39:00 EDT

I found a scanned note on the Internet earlier today.

This reminded me of something from my personal education history.

Rewind way back to elementary school. I was in fourth grade. I had a science class, wherein the science teacher asserted that air did not have weight.

He proceeded to demonstrate this with an experiment. He took two balloons, filled both with air, and tied them both to a ruler. The ruler was suspended by a string from the halfway point. It didn’t quite sit level, largely because it was leaning against a shelving unit, rather than being free-standing. He than popped one of the balloons. Since it still wasn’t quite level, it was concluded that air does not have weight.

I started to point out that the experiment was flawed. The ruler wasn’t level to begin with. When the balloon was popped, some bits of balloon fell off, which would affect the weight on that end. He cut me off, saying something to the effect of “well, the experiment isn’t perfect, but air still doesn’t weigh anything.” The class agreed (the teacher must be right!) and wanted to move on.

Then I asked him how barometers worked.

“Disregard for authority” indeed.

EOS 400D: Rocks!

Tue, 26 Sep 2006 17:54:00 EDT

As anyone who hasn’t been following my Flickr account knows, I received my Digital Rebel XTi last Monday. It’s great. A few highlights, from my recent visit to Windley Key state park:

T-Shirts

Sat, 16 Sep 2006 21:15:00 EDT

We just ordered two new shirts for Sarah. Check it out:

I thought they were fun.

EOS 400D: Monday

Sat, 16 Sep 2006 20:04:00 EDT

My Digital Rebel XTi was shipped on Friday. Yes!

History Lesson

Let’s go way back in time. Canon sells a lot of cameras—not just the point-and-shoot variety, either. They’ve had a set of “pro” equipment built around a common lens system since the early seventies, and due to a variety of design constraints, decided they needed to break compatibility. They dropped the FD lens mount, and introduced EF.

EF is a step up from FD. With FD, the camera mechanically moved things inside the lens to focus and such; EF stands for electro-focus, and as the name implies, EF lenses focus themselves electronically. The camera mount has no moving parts, just a series of electrical contacts. This new lens mount, compatible lenses, and compatible camera bodies were collectively bundled into a new SLR line called “EOS”, born in the late eighties.

In the late nineties, Canon realized that digital photography was starting to become practical for some professional applications, and started work in making an EOS camera with a digital sensor instead of film. In 2000, the D30 was born—fully compatible with the EF lenses that many photographers already owned, which was good, considering the $3000 price tag.

The D30 also marked the introduction of Canon’s CMOS digital sensor. Unlike other digital SLRs (including Nikon and Sony), Canon develops their own sensors in-house. Also, unlike other digital SLRs, Canon uses a CMOS sensor instead of a CCD. While CCDs are capable of higher efficiency on paper, CMOS has a number of real-world advantages, including the ability to build an image sensor and support circuitry into different layers of the same piece of silicon. This means that Canon can go from light-sensitive pixel to noise reduction logic to image transfer backplane in a few microns of silicon, instead of needing to build all of that on the periphery of the device (or worse).

Time passes. Canon announces a digital EOS-1 (their highest-end model), upgrades from the D30 to the D60 to the 10D, and eventually Canon decides to produce a sub-$1000 digital SLR. Thus was the Digital Rebel.

More time passes. The 1D cameras get upgraded, the 20D replaces the 10D, the Digital Rebel XT replaces the Digital Rebel, the 5D is added, the 30D replaces the 20D, and just last month, the Digital Rebel XTi replaces the Digital Rebel XT. Score!

Analysis

For anyone that got confused, please reference the table below.

Line	Camera	Year	Resolution	Price
Full-frame pro	1Ds mark II	2004	16.6 megapixels	$7,000
High-speed pro	1D mark II N	2005	8.2 megapixels	$3,550
Full-frame	5D	2005	12.8 megapixels	$3,000
Semi-pro	30D	2006	8.2 megapixels	$1,200
Consumer	Rebel XTi	2006	10.1 megapixels	$800

Sidenote: all of the cameras except the 1Ds and the 5D have a so-called “crop factor” of 1.6. This is because the sensor used in place of the film is smaller than 35 mm—they’re 22.5×15.0 mm. So, if you use a lens designed for a 35 mm plate, you’ll only capture 22.5 mm, effectively zooming in by a factor of 1.6. This means that a 50 mm lens on a XTi or a 30D gives the same view that an 80 mm lens on a 35 mm camera would. Anyway.

Notice anything from the table? Compare the XTi to the 30D. They’re basically equivalent, except the 30D takes 5 frames per second (compared to the XTi’s 3), can go to ISO 3200 (compared to 1600), and the shutter can go down to 1/8000 of a second (compared to 1/4000)... however, the XTi captures at a higher resolution, has a unique self-cleaning sensor, is 30% lighter, and is $400 cheaper. Cool, huh?

This is a unique position for Canon. When they announced the Digital Rebel, it was clearly the bargain-basement digital SLR. The camera body sold for $899, and in order to call it a “sub-$1000 digital SLR”, Canon needed a $100 lens. They came up with the EF-S 18-55mm f/3.5-5.6 lens. Compare to the 20D and 30D, which sold either body-only or with the EF-S 17-85 f/4.0-5.6 IS USM lens, or the 5D and 1D/1Ds that never sold with a lens.

Application

Why does this matter? Seems that Canon recognizes that the XTi will cannibalize their 30D sales. They’re forced to give it those specs in order to compete in the sub-$1000 market. In response, they’re selling the camera body-only for $799 under part number 1236B002, with the 18-55mm lens for $899 under part number 1236B001, or (like the 20D and 30D) with the 17-85mm lens under part number 1236B006. However, if you look, you’ll find that no one actually has the 17-85 kit, and that many Canon dealers haven’t even heard of it. However, some have.

Anyway, I wanted the 17-85 mm lens for the wider range of focal lengths, the third-generation IS, and the ring-type USM (more on those later). Despite various other retailers getting the camera early this week (or even late last week), I decided to hang out until Norman Camera got a shipment of black bodies. Why? They would sell me a 1236B002 and the lens in separate boxes, but ring it up as a 1236B006 for $100 less than everyone else, and $150 less than buying both separately. Score!

Conclusion

So, if you want the XTi and a 17-85 IS USM, call Norman Camera. They’re an authorized dealer, selling legitimate products, with good ratings from everyone that rates stores. And you’ll save a lot of money. Also, others that offer cheap cameras try to charge you extra the charger and the battery that come with the camera. Norman does not.

Happy surfing!

9/11

Mon, 11 Sep 2006 15:33:00 EDT

Cox and Forkum publish political cartoons five times a week. Every year, they sum up what’s happened since last year by adding a panel to a series entitled “Confronting Terrorism”. I rearranged the panels chronologically, figuring that it’s easier for the uninitiated to understand them this way.

See also their previous September 11 editions in 2004 and 2005.

Smoking is Good For You

Mon, 11 Sep 2006 14:59:00 EDT

Or, at least, these people would have you believe so. They say things like “in short, the “dangers” of passive smoking are a fraud”, assert that lung cancer affects only 3% of smokers, and so on.

Their FAQ says that “to re-establish common sense and balance, it is necessary to oppose healthism extremism with an equal counter-force”—apparently, this equal counter-force includes other snippets like:

Wait till the transport service gets wind of that. They could save a lot of paperwork by firing all their drivers right now. Give the bus keys to Don Ward. Don could establish a route running by Mike Bloomberg’s, Ev Koop’s, and Henry Waxman’s houses. The unemployed drivers, can join the rest of us vulgar stinkers, in migration to Belize. Things aren’t quite so “eccentric” in Belize.

The state of Massachusetts is no better than the Nazi regime it bases such laws on.

Chop off all students’ hands. One can’t flick a BIC, let alone strike a match with stumps. Cruel to be kind? Maybe, but those little bastards must be saved from smoking at any cost.

Their de-bunking of studies leaves some room for improvement, too.

SECONDHAND SMOKE MAY CAUSE CAVITIES IN CHILDREN! – Of all the absolute nonsense about smoking, this “study” takes the prize, and its well-earned place in this Theatre. The study simply “forgot” to account for was diet and dental hygiene habits (tooth brushing), but what the hell… as long as it helps to make smokers look bad, anything goes… The corruption of science to a political agenda cannot be more obvious.

The article (which they link to, by the way) goes on to say that “this relationship between cavities and second-hand smoking persisted after we controlled for many variables, including age, sex, race, region, dentist’s visits, nutritional status and blood lead levels.” I would think that “dentist’s visits” and “nutritional status” addresses the diet and dental hygiene questions. Additionally, it’s about 4,000 kids, which is certainly large enough to derive a statistically meaningful data between smoke and cavities, even with that many controls.

Good times.