house-buying recommendations

I bought a house about a month and a half ago. It's been a pretty interesting experience, and some people have been asking me for recommendations, so here's what I did:

In October I'd done a bit of looking around on Zillow to get an idea of what was around. I'd already had a price range in mind (around $175-200K), an idea of where I wanted to be (in north Austin, not too far from work, and particularly not where I'd need to take a toll road to get to work), and a rough idea of what I wanted in terms of features (single story, two-car garage, gas hookups for range/dryer, etc).

I also went to get pre-approved for a mortgage. I didn't really do a lot of shopping around here, I just went to talk to Matt Schepper at University Federal Credit Union, where I've been a member since college. My credit is pretty much nonexistent (never had a credit card, never had an auto loan), so I did have to bring in quite a bit of paperwork to show that I had quite a bit of capital available, but other than that it was pretty smooth.

Eventually you do hit a point where just browsing Zillow isn't quite enough, and you need to get a realtor, so in November, I got a recommendation for one from @schlaeton. Jeff Kress is very familiar with the north Austin/Cedar Park area, and I would recommend him to anyone else.

I sent Jeff my requirements and a list of the houses on Zillow that I had on my mind, he provided a few ideas of his own and put me on some MLS system to notify me via e-mail with new house listings, and then on Tuesday and Friday on Thanksgiving week we actually went and looked at houses. The speed at which I could do a go/no-go on each house was a bit surprising to Jeff, but I'd already had a good idea of what I wanted, and, being single, I didn't have to argue my choices with anyone but myself.

So, of the dozen or so houses in total, I very quickly narrowed it down to three, then to one. We put an offer on it, and after a couple rounds of negotiation it got accepted.

For inspection I took @kcipp's recommendation of Precision Inspection. Ronnie White was the inspector; he seemed a little grumpy, but at the conclusion he gave me a rundown of everything and how difficult some of those things are to fix, and was pretty thorough in his report.

For insurance, I tried working with a State Farm agent at my realtor's suggestion, but the prices I was quoted from them came out to be about double what I was being quoted for through Progressive, with whom I have my auto insurance, so I went with them instead.

Also, as part of the sale, the sellers pay for a year of a home warranty with a vendor of my choice. I went with American Home Shield, since that's what a couple people I asked were using (@fearthecute and @yardkgnome if I remember correctly). I haven't had to call them up at all, so right now I'm pretty ambivalent about them.

I closed the first week of January, which means that I don't have to deal with all the tax implications quite yet. I've already made a couple IKEA trips, because I barely had any furniture. I still need to get living room couches at some point. And a washer. And probably some way to mow the backyard. And get started on the dozen or so different house projects I have in mind already...

on backups

Oh, hey, I'm blogging again. We'll see how long this lasts.

SSDs have been coming down in price, and I've been thinking about hopping on that train. One of the biggest concerns about SSDs, though, is their lifetime, which is why I've been trying to come up with a solid plan for backups.

Yes, a solid plan for backups. I've never made one. Somehow over the course of two decades or so I've not been bitten in the ass by that.

Of course, I don't have a plan for backups is not the same thing as I've never made backups-- I generally make sure that any data I have exists in more than one place (in practice, on more than one computer) at a time, but my means for doing so have always been ad hoc and never on any sort of regularity.

My current basic thought is something along the lines of:

  • My desktop rsyncs an 'Archived' folder to a Linux fileserver I keep at home, at some interval.
  • The fileserver rsyncs the same folder to my Linode VPS, at some interval.
  • At some interval I make snapshots to some more-permanent means of storage (DVD, tape, CF cards, no idea).

This approach would give me both on-site and off-site backups, which would be nice. And the off-site backups go to the motherfuckin' cloud, which is all full of rainbows and unicorns and shit, yo.

But thinking about this a little more, that really only addresses the problem of one computer's backups. There's some of that data that I would like to have accessible from multiple computers, and I'd really like to be able to handle that in a way better then schlepping eight gigabytes of data around everywhere. Really, the use cases become:

  • I want all my data to be backed up in both on-site and off-site means.
  • I want to not be dependent on a specific third-party service. I can always migrate software on my VPS to another VPS provider. I'm a little more screwed if I'm depending on Picasa and Google pulls a Wave.
  • I want to be able to access documents from both my desktop and my laptop.
  • I want to be able to get to my photographs from my desktop, laptop, and my phone.
  • I want to be able to do these things without necessarily having downloaded them to those devices a priori.

I think really what I'm after, in that case, is closer to Dropbox, but I'd need to shell out money for their 'Pro' service for the amount of space I'd need, and if Dropbox goes away then I'm back at square one.

SparkleShare might be worth a try, perhaps, but its use of git as a backend has me concerned about performance with large binary files (in my case, mostly photographs). Although in this case, I guess they won't be changing much, so maybe it's alright. It doesn't address some of my other use cases (like being able to get photos on my phone) though.

latest project idea: beaglebones and lcds

Recently we've been tossing around a product idea at work. Some of the possible designs called for a USB gadget driver, so in order to experiment a bit with GadgetFS I bought a BeagleBone.

Actually. That makes it sound so simple. Actually, what I did first was take a look at a bunch of different ARM boards (guided in part by Linaro's list, and in part by some of the ones we have at work), compare them, and choose:

  Pandaboard ES Pandaboard BeagleBoard-xM BeagleBone i.MX53 Quick Start
Cost $185 $174 $149 $89 $149
Processor Dual-core Cortex-A9 @1.2GHz Dual-core Cortex-A9 @1GHz Cortex-A8 @1GHz Cortex-A8 @720MHz Cortex-A8 @1GHz
Memory 1GiB 1GiB 512MiB 256MiB 1GiB
Networking 10/100 Ethernet, 802.11bgn, Bluetooth 10/100 Ethernet, 802.11bgn, Bluetooth 10/100 built into USB hub chip 10/100 via on-chip controller 10/100
USB 1xOTG, 2xHost 1xOTG, 2xHost 1xOTG, 4xHost 1xOTG, 1xHost 1xDevice, 2xHost
Video HDMI+DVI HDMI+DVI DVI-D (HDMI connector) None VGA

This is by no means an exhaustive list. (Especially with the recent announcement of the Zynq-based ZedBoard, which has a combination Dual Cortex A9 processor and FPGA. Drool. Although it's in a different space, I suppose.)

I was pretty tempted to go all the way and pick up a Pandaboard ES, but I don't really need any of the additional features and it'd just be overkill.

One thing that was a little sad about the BeagleBone was the lack of a video output device. Kind of. The OMAP that powers it has an LVDS interface off of the processor, so now I've got it in my head to try to build an interface from my BeagleBone to a laptop LCD scavenged from my old dead Sony Vaio VGN-S150. There should be an LVDS interface going from the laptop to the LCD, so it should be easy to just swap in my BeagleBone to be the 'laptop', right?

Plus it'd be far more interesting then just buying a DVI-D cape.

One of the unfortunate things about laptop LCDs is that, while they abide by similar electrical standards, there aren't many physical standards. Identifying my LCD panel was as easy as looking at the sticker on the back and finding it's a Toshiba Matsushita LTD133EX2X. So let's turn to Google to see what we can find about it:

Google search results for said panel

All that lead me to were crappy SEO-optimized websites that just cataloged parts, all the while claiming that they had datasheets but really had nothing at all. What a waste of bandwidth.

Using some other queries I was able to dig up this collection of video standards, which includes the VESA Notebook Panel Standard v1.0 as well as the Standard Panels Working Group website, which prescribe (the same) recommended pinout. It doesn't really match.

So then, I thought, maybe there are similar panels that Toshiba Matsushita made at around the same time that would have the same pinout? I used one of those crappy SEO-optimized don't-really-have-a-datasheet websites to get a list of other panels, and then Googled around to find datasheet for the LTD133EX2A, another 1280x800 6-bit panel.

So now I have two possibilities for pinouts:

Pin Current wiring VESA/SPWG 6-bit Panel LTD133EX2A
1 Thick Blue GND: Ground Vdd: +3.3V
2 Thick Blue AVDD: 3.3V Vdd: +3.3V
3 Thick Black AVDD: 3.3V Vss: GND
4 Thick Black DVDD: DDC 3.3V Vss: GND
5 Black (in twisted pair with 6) No Connect RxlN0-: Neg LVDS data input (R0-R5, G0)
6 Green (in twisted pair with 5) SCL: DDC Clock RxlN0+: Pos LVDS data input (R0-R5, G0)
7 Grey SDA: DDC Data Vss: GND
8 Black (in twisted pair with 9) RinX0-: Neg LVDS data input (R0-R5, G0) RxlN1-: Neg LVDS data input (G1-G5, B0-B1)
9 Yellow (in twisted pair with 8) RinX0+: Pos LVDS data input (R0-R5, G0) RxlN1+: Pos LVDS data input (G1-G5, B0-B1)
10 Tan (twisted with 11 and 12) GND: Ground Vss: GND
11 Black (twisted with 10 and 12) RinX1-: Neg LVDS data input (G1-G5, B0-B1) RxlN2-: Neg LVDS data input (B2-B5, HS, VS, DE)
12 Brown (twisted with 10 and 11) RinX1+: Pos LVDS data input (G1-G5, B0-B1) RxlN2+: Pos LVDS data input (B2-B5, HS, VS, DE)
13 Yellow (twisted with 14, 15, and 16) GND: Ground Vss: GND
14 Grey (twisted with 13, 15, and 16) RinX2-: Neg LVDS data input (B2-B5, HS, VS, DE) CLK-: Neg LVDS clock input
15 Tan (twisted with 13, 14, and 16) RinX2+: Pos LVDS data input (B2-B5, HS, VS, DE) CLK+: Pos LVDS clock input
16 Brown (twisted with 13, 14, and 15) GND: Ground Vss: GND
17 not connected CLK-: Neg LVDS clock input No Connect
18 not connected CLK+: Pos LVDS clock input No Connect
19 Green GND: Ground No Connect
20 Grey GND: Ground No Connect

Unfortunately, neither is perfect, but given the placement of the power wires I really feel like the LTD133EX2A pinout is closest; although there is the weirdness with pins 19 and 20. Perhaps the LTD133EX2X added DDC information and those are the Clock/Data lines? I'll have to throw my Bus Pirate at it and see if I can get anything over I²C.

And then, of course, once I figure that out, then I actually have to do the interfacing. I seem to have two options here:

I may actually need to learn how to do SMD soldering for this project. I guess I've managed to put it off for long enough.

And that's all completely aside from the GadgetFS stuff that lead me to buy this board in the first place.

apparently the rnc hasn't caught on yet

I got another 2012 Presidential Platform Survey from the Republican National Committee. So what's different this time?

The intro letter is worded pretty much the same, with some minor edits now that Romney is the presumptive nominee. Some examples:

  • 'these and other issues that are dominating the 2012 Presidential campaign!' becomes 'the issues dominating the campaign debate between Mitt Romney and Barack Obama!'
  • 'electing Republican leaders' becomes 'electing Mitt Romney and Republican leaders' (which I suppose may be a backhanded statement on how capable of a leader the RNC views Romney)
  • A couple paragraphs talking about what the RNC is and does got reworded and rearranged.

Curiously the margins on the intro letter seem to be ever so slightly different; some of the paragraphs are word-for-word identical, yet they've flowed on the page differently.

As far as the survey itself, I see the following changes:

  • There's now a demographic question asking if I plan to volunteer for the local Republican Victory Center for this election.
  • 'Defending the Constitution' is no longer an item under the 'how important is it to voters that candidates give attention to these issues' question in Section I
  • The loss of a question in Section II asking if Congress should block efforts to raise the debt ceiling.
  • The loss of a question in Section III asking if I support reform of how the government pays for Medicare for future retirees.
  • A new question in Section IV about whether Obama's decision to eliminate the military's "two war" strategy.
  • The loss of a question in Section IV asking if it's time to withdraw from Afghanistan.
  • A new question in Section VII which really amuses me: 'Are Republicans in your area enthusiastic and committed to voting for our Republican slate of candidates this November?'

Of course, they haven't let up in their attempts to try and extract at least $15 from me, for 'tabulating my Survey'.

klout is useless

A while back, I signed up for Klout for shits and giggles (this was before I found out that they were fans of bro'ing down and crushing code).

What is Klout? 'Klout measures your influence on your social networks.' Basically, they have some sort of automated process that tries to figure out, based on Twitter, Facebook, LinkedIn, and other social networks, how influential you are in various topics. So what topics do they think I influence?

  • Apps: I'm a driver developer.
  • Books: I don't read much.
  • iPhone: Haven't owned one for about a year and a half now.
  • Business: I'm terrible at it.
  • Camera: I have one on my phone, just like millions of other people.
  • Gaming: Okay, it got this one right.
  • Creativity: I'm not sure what this one is supposed to be. There was a period of time where I was getting back into drawing but I haven't in a while.
  • Seattle: I've never been there.
  • Tools: I'm not sure what this one is supposed to mean.
  • iOS: The only iOS device I ever owned was the iPhone I got rid of a year and a half ago.
  • Job Search: I've been working at the same place for about four years.
  • Yoga: I don't do it.
  • Tb: I can't tell if this is supposed to be tuberculosis or Thunderbolt. Either way, I'm not very influential in either.
  • Debt: I'm not in it.
  • Marriage: I'm single.
  • Austin: Finally, something that fits: I live here.
  • Beer: This fits too.
  • Texas: I also live here.
  • Law: ... and we're back to ones that are irrelevant.
  • Traffic: I guess I complain about it a lot.

I figure that's about 25% accuracy on a topic-by-topic basis (Gaming, Austin, Beer, Texas, Traffic), and what's worse, four of the ones that I consider potentially relevant are things that Klout has marked as 'low'.

And yet people are basing jobs on this shit. Sigh.

i would make for a very poor republican

Today I received a 2012 Presidential Platform Survey in the mail from the Republican National Committee.

I scanned it to share (1.6MiB PDF), of course. The actual survey portion is on paper about 10.75"x14", which made it a bit difficult to scan.

The enclosed letter states:

After compiling and modeling demographic information for the thousands of activists in our database, you were selected to represent voters in your area in the OFFICIAL 2012 Presidential Platform Survey.

Enclosed is a special questionnaire document - REGISTERED to your name and address -- identifying you as THE DESIGNATED REPRESENTATIVE of VOTERS residing in your district.

It seems that if they 'compiled and modeled demographic information' for 'thousands of activists' they would have managed to pick someone who is, I dunno, actually somewhat enthusiastic about the Republican party? Maybe this is karma for me having voted in the Republican primaries in 2010?

The letter also strongly implies that I can't throw it out, that it must be tabulated in my name,, etc. I seem to recall that anything that is mailed to me unsolicited becomes my property, but I can't find the appropriate laws on the subject. Something about companies sending stuff and then sending invoices, or something.

I also enjoy how the letter highlights that Obama is 'campaigning full-time' (since I guess he's solved everything so now the presidency is just a walk in the park these days) and that he's 'raising tens of millions of dollars' (unlike the combined $205.6 million from the nine Republican candidates as of the end of February).

The questions are... wow. This post by Don Nordeen, who also received one of these surveys, sums up my feelings about it. Interestingly, my survey appears to be slightly different: mine does not have a question about potential support of another bailout, but rather there's a question about concern about inflation increasing my cost of living. He also has a question about medical malpractice reform that I do not.

Some of the questions, especially toward the end, are obvious baiting. It's clear that they're looking for one response and one response only.

30. Do you believe the Republican Party needs to do a better job of exposing the Obama record and his radical liberal agenda?

31. Are you committed to helping ensure that in 2012, the Obama-era of radical liberalism, reckless spending and embarrassing foreign policy comes to an end?

In the end, it's a survey that won't actually tell the Republican party anything but will just serve to validate their positions. The last thing they need is an echo chamber.

After completing the survey, it would seem that I would be left with three options in the 'Support Reply Form' section:

  • YES!, I want to help elect Republicans all over the place, here's $35-500.
  • No, I don't wish to participate, but here's $35-500 anyway. (I guess this is if you want to donate but you don't want your survey results counted, or something?)
  • No, I don't wish to participate, but here's $15 to cover 'tabulating my survey'. Seriously. What the hell?

So I'm left wondering what I should do with this survey. I could simply ignore it (despite the stern warnings, the Republican National Committee is no Census Bureau). I could send it back blank. I could fill it out, either as a deep-end right-wing (giving them the answers that they clearly expect) or as myself (which would be the opposite choice for nearly all of the questions), but then I'd have to give the RNC $15. Or maybe I could just send it without money. Or send it with 11.24 EUR, perhaps? I'm sure that'd piss somebody off.

on android and compatibility

I've been looking into game development via the Android NDK.

Both the Android SDK and NDK are divided into API Levels; by targeting a particular API level, you get access to all features from that version of Android on, but your application won't run on earlier versions of Android.

API Level 9 (for Android 2.3) provides a couple of features that I'm interested in:

  • Native Activities, which allow you to write activities purely in C or C++, without a single line of Java. This is attractive from a maintainability standpoint (less languages to deal with, less complicated build process, etc) and because most of our game logic will be C++ for portability anyway.
  • OpenSL ES 1.0 support, for being able to do audio natively from C++ without having to make JNI calls back into the Java runtime to talk to the SoundPool APIs.

Restricting ourselves to Android 2.3 and later is also attractive from a testing standpoint; I have two Android devices that I can test with (a T-Mobile G2 currently running 2.3.7, and a Nook Color currently running 4.0), so I'd have to find an older device for testing.

The question then becomes: How many Android handsets will that cut us out of? To answer that, I tried searching the web for published analytics on Android versions before finding that Google actually provides helpful stats from Google Play logins, which is probably the most representative sample set. The bad news: At least as of this writing, Android 2.2 (Eclair) and earlier still comprise 33.1% of the market.

Stacked line chart indicating Android version breakdown between September 2011 and March 2012

Of course, that's today. Several months from now the landscape could look rather different: a rough extrapolation would suggest that that number goes down to about 12% in another six months... assuming that trends remain the same.

So why are people still using these handsets? It's hard to say, and without more data I don't really know.

Android 2.3 was released in December 2010, 26 months ago. It takes some time for handset manufacturers and carriers to begin deploying 2.3 to hardware (a problem that Apple fans love to point out), but many phones do get upgrades. I looked at Wikipedia's List of Android HTC phones to get a good idea of timelines. It notes that the newest HTC phone to be stuck on 2.2 officially (i.e., not taking CyanogenMod into consideration) is the HTC Gratia, released Nov 2010, which was by-and-large a rebranded HTC Aria, released Aug 2010.

But this brings to mind two questions:

  • Are the people buying these phones on the standard American two-year upgrade cycle? Should I expect them to trade up to something running 2.3 or 4.0 by the end of this year?
  • How well did these phones sell? More specifically, what is the device breakdown for these operating systems?

I can't find any data to answer the second question, which is unfortunate, because I'm not only concerned about the level of OS support, but also the capabilities of the hardware; if these are devices with slow CPUs or if they can't do OpenGL ES 2.0 or have low memory, then maybe I shouldn't worry about targeting them. Maybe. I don't have the data to know for sure. I would love if Google published statistics as detailed as the Steam Hardware Survey but sadly they don't.

ipod mini

Yes, I still have one of these. I replaced the 4GB CompactFlash microdrive with a 16GB solid-state CF card, and the battery life is pretty much shot, but it works well enough for car audio.

san francisco pier

From a photograph I took on my phone at one of San Francisco's piers.

gdc trip: day five

And all good things must come to an end.

Romer and I were able to meet up with @VTPG for breakfast at Joe's 24th Street Cafe before meeting up with Trevor, Peter, and Stephen to head to the airport together.

On the SFO-to-DFW leg, I had a woman sitting next to me clearly talking to her boyfriend in that sort of lovey-dovey I-can't-go-without-being-in-constant-contact-with-you way. Fortunately, our flight did not have internet access, much to her disappointment as it meant she couldn't do a Skype call with him mid-flight. Never before have I been so glad to not have internet access available.

At DFW our flight to AUS was slightly delayed because nobody had cleaned the plane. I'm not sure what was up with that.

We did observe a curiosity about airline boarding though. Airlines will try to board all the 'preferred' classes first, so that they can get seats in front of the plane... which means that the entire line is held up while they get into their seats. What airlines should do is seat coach first-- yes, it means that first class has to wait a little longer to board, but seats are assigned already so what does it matter? They're already getting more room, and they get to deboard first. Getting everyone set up on the plane faster means that they can get in the air faster.