Rhyme without reason...

I saw this on the WeatherBug website, and thought it was a pretty creative way to sell fire logs (click on it, then wait for it to load):

This is just a quick note to hopefully save others the two hours of troubleshooting I just went through.

If you are using htscanner (currently version 0.9.1-dev here), do not use bit value suffixes when overriding php.ini directives in .htaccess files; you have to write the values out as bytes without a suffix.

For example, when specifying a setting like "post_max_size" in either php.ini or in the .htaccess file, PHP normally allows you to specify something like 16M for 16 megabytes. Htscanner, on the other hand, requires you to write it out in bytes, without a suffix.

The most frustrating part is that htscanner will still parse it, and phpinfo() will show that the setting took effect as 16M, but PHP is actually interpreting the value as 16 bytes.

Here in the OPL, we ran into this issue on Drupal 6 when we increased the post_max_size setting in our .htaccess file, but only when using the FileField and Upload modules. Basically, we were seeing that any node form submissions were being ignored, or were getting the dreaded "white page of death". After some digging, we also found that any PHP form with an encoding of multipart/form-data was being affected, such that the $_POST super-global was entirely empty. Digging even further, we found that if we set post_max_size in bytes, rather than with the suffix, the problem went away. Discovering that htscanner was doing the actual processing of the value, and that it didn't support the suffixes, explained why.

The following bug report makes mention of this odd limitation in htscanner:
http://pecl.php.net/bugs/bug.php?id=16130

So, after dealing with several RIT departments today that all encountered issues their current process didn't seem to anticipate, I got an idea for a tacky, yet hilarious YouTube video that I wish I could create, but I don't have the resources for.

The script goes something like this:

An overly-serious narrator (similar to the guy from the Power Thirst commercials) says "Process is what defines an organization. Does your company have what it takes?". The commercial then cuts to a scene of a customer service representative at a counter, with people waiting in line.

Customer Service Representative: Hello sir, how can I help you?

(The man walks up to the counter)

Man: Well, I was here yesterday to get the copies of this brochure printed, but it looks like some of the information they printed was wrong.
Customer Service Representative: I'm sorry to hear that, sir. Let me see a copy.

(The man hands the representative a copy)

Man: See? The company name is wrong.
Customer Service Representative: Ah, I see. Well, we just write in what we're given in the order.
Man: Can I see the order information?
Customer Service Representative: We had the sheet for it yesterday... let me see...

(The Customer Service Representative rummages around in a recycling bin nearby)

Customer Service Representative: I'm sorry, but we don't have it. I think... we shred the order slips after processing the order.
Man: You don't file them in case there's a problem later?
Customer Service Representative: No... we don't.

(The Customer Service Representative suddenly looks concerned)

(The scene freezes with focus on the Customer Service Representative's concerned expression, and the text "Process FAIL" comes on screen in big, red writing)

Narrator: She just found out her process FAILED.

The video goes on with similarly absurd scenarios of people running into cases that their processes didn't anticipate, each scenario ending with the narrator saying "HIS process just FAILED", "HER process just failed", "THEIR process just FAILED", etc.

The video then ends with "Process. Is your organization ready?"

I could easily see it being a hit in Software Engineering classes.

A lengthy upgrade process; an inexplicable failure to boot after install; slowness after install; hardware drivers that don't load properly; false claims of improved performance and reliability; software that used to work but no longer does; sparse support documentation, and support staff that seem unprepared to deal with the issues... what OS am I describing?

Windows Vista? Well, sure, Vista matched those criteria perfectly in its first year and a half, but it's not the one I'm describing this time.

Windows 7? No, certainly not – Microsoft seems to have learned their lessons from Vista's failures (at least for now) and Windows 7 is a spectacular, stable, well-thought-out OS.

No, the OS I'm describing is from a different company.

Apple, welcome to your Vista – Snow Leopard.

Last night, my fiancé did an upgrade install from Leopard to Snow Leopard. The whole process took about 45 minutes, and afterwards, what greeted his awaiting eyes? Why, the Apple logo and the startup spinner... for over an hour. As it turned out, the installer botched the upgrade and now the system refused to boot.

Booting into single-user mode only revealed that the system was getting stuck after enabling Intel CPU power management, but there was not a single error in sight (which is typical for Apple products; you can compare that to Microsoft's approach of giving you an unhelpful error, but at least it gives you something definitive to Google).

When we tried booting off the Snow Leopard DVD, the system seemed like it was booting and then the screen turned black. It seemed like the system had shut off, but it wasn't responding to the Power button. As it turned out, the Apple backlight driver was getting some bad data from Parameter RAM (PRAM), of all places – which seems to be a common problem for many users trying to install Snow Leopard. After 10 minutes of locating things on the screen with a flashlight (the proverbial "spotlight"?), we ended up just clearing PRAM (Command-Option-P-R) which fixed that issue, but I haven't had to do that procedure in a very long time (like, PowerBook 1400 era). We re-ran the install, and 55 minutes later the system was bootable... sort of.

After boot, the system rebooted once, and then it took about 1 minute and 30 seconds for the log-in window to open. About 2 minutes and 30 seconds later and he was at his desktop, to find that the OS had forgotten his file associations for the Adobe Creative Suite. With the new OS, it now takes longer than a minute to launch any program. Why, this OS certainly does have an astounding effect on performance, just not the kind Apple had advertised.

Apple fans may be quick to jump to Apple's defense:

• But, but, upgrade installs are always risky! Windows upgrades are just as risky as Mac OS X upgrades, and usually fail more often! Plus, Windows upgrades are often just as slow afterwards!
  
  That's true, but consider that Microsoft has to deal with a huge diversity of hardware. Most failed Windows upgrades fail because of third-party drivers and programs that hook into the kernel and core of the previous OS that don't load correctly in the new one. Apple makes the hardware and knows every system configuration they're supporting. Furthermore, there are far fewer applications that hook into the OS on OS X.
  
  Second, Apple's marketing material touts that users can easily upgrade from Leopard and re-claim disk space. It seems like they're encouraging users to do an upgrade install, and there is nothing in the documentation that suggests that an upgrade install is more risky.



• Well... your system is just an isolated case!
  
  A quick Google search would suggest otherwise. Again, his system isn't running anything special and its all stock Apple hardware, so the blame still rests on Apple's shoulders for encountering these issues.

The moral of the story is this: Apple could have learned from Microsoft's, Intel's, and nVidia's mistakes with Vista, but they didn't. Instead, they sat on the sidelines and poked fun at Microsoft for their own marketing gains.

Instead, Apple repeated the same mistakes:

Vista	Snow Leopard
The primary "feature" of Windows Vista was an improved security model, but the majority of changes it brought were under the hood – a better deployment model for IT staff, better encryption, a better DirectX API, etc. Unfortunately, it's hard to sell users on under-the-hood changes because they're things the users can't see or touch directly.	Snow Leopard has no real new features for the user, except for Exchange support (which only helps corporate customers) and a few tweaks here and there. Most of the changes are under the hood – supposedly improved performance, a new threading API for developers to optimize their applications for multiple cores, reduced bandwidth usage for iChat, GPU-accelerated video decoding support in QuickTime, etc. Essentially, Apple is trying to sell people on under-the-hood changes.
In scrambling together the drivers for Windows Vista, nVidia made significant changes to their video drivers that supposedly improved performance, but in the rush to get the drivers out, they neglected to test them heavily with some of their older products. As a result, many users who still had nVidia's mid-range video cards from a few years back found themselves with crashes and failed boots in both Vista and XP. Had Microsoft and nVidia taken some time out to test these drivers more thoroughly rather than rushing to get them out more quickly and save face, they would have ended up looking better.	Apple released Snow Leopard earlier than users expected, no doubt to get it out earlier than Windows 7's official release in October in some "clever" marketing move. Unfortunately, had Apple learned from nVidia's mistakes, they would have realized that performance tweaks often require more testing, especially on the older hardware (2007 era) that they still supposedly support. In the end, they would look better to the hundreds of users like us that have run into such embarrassing problems with the install.

We'll ride this one out. Unfortunately, unlike Vista, we have no "Service Pack 1" we can look forward to that will fix these issues, and Apple is often far less willing to admit its' errors than Microsoft.

The marketing material for Snow Leopard says its "the world's most advanced operating system", and borrowing from Steven Wright, "Snow Leopard is so advanced, you don't even need it."

If you find yourself having to work with large numbers in PHP, you'll more than likely need GMP library support.

Unfortunately, if you're running Debian Etch (Debian 4.0), "php5-gmp" – the package that provides GMP library support for PHP – is not officially available, and the version that's available for Lenny (Debian 5.0) requires "php5-common" to be exactly version "5.2.6.dfsg.1-1+lenny3", which would break a lot of other packages (including Apache) that depend on php5-common.

To make matters worse, after browsing around the web, you might come across an article that says it provides php5-gmp for Etch, but when you try to go to the download location, you get a 404 "File Not Found". Even worse, when you download the package from yet another page, the package you get depends on exactly version "5.2.0-8+etch11" of php5-common, but "5.2.0+dfsg-8+etch15" is the latest secure version currently out for Etch. What a mess!

Luckily, I've done the hard work of unpacking that last package, adjusting the dependency requirements, and re-packaging it so that php5-common version "5.2.0-8+etch11" or later is required, allowing the later version in Etch to satisfy the dependency. You can download the result here:

http://cias-opl.rit.edu/misc/php5-gmp.deb

To install it, use "dpkg -i".

Please note that this practice of adjusting dependencies and re-packaging is discouraged, especially for packages that are closely linked (like php5-common and php5-gmp) as a linkage difference between the two could lead to crashes or unexpected results. Consequently, your mileage may vary, but for me this tweaked package seems to be happy and provides the needed functions. And APT is happy as well, which is always good.

Back in April, I wrote about how users of the Linksys WUSB54GP could get a much more stable wireless connection by swapping out their drivers for those of the D-Link DWL-G122 Rev. A Wireless USB adapter. That was all well and good for Windows XP, but when I switched into Windows Server 2008 and then to Windows 7, I found that the D-Link drivers had one fatal flaw: at shutdown time, the OS would come to a screeching halt with BSOD and STOP code 7F (double-falt).

Although it was possible to avoid the BSOD by disabling the wireless adapter before shutdown, this seemed awfully tedious, and I found myself often forgetting to do it before clicking the reboot button on the start menu. Desperate for a solution and unwilling to go back to the mediocre-at-best Linksys drivers, I again headed over to the prism54 project's website to look at the list of supported devices with the same chipset as my current adapter. As before, my goal was to find find yet another device with compatible drivers that did not have the crash on shutdown.

After trying about 5 different driver sets, I discovered that the "IOGEAR Wireless-G to USB 2.0 Flex Adapter (GWU513)" had exactly the drivers I was looking for. They can be downloaded here:
http://www.iogear.com/support/dm/driver/GWU513

The tricky part is that they bundle the drivers inside an installer that can't be extracted with the Universal Extractor, and the installer wants to perform the driver update itself. If it fails to update the device with the drivers (which it will do because it doesn't recognize the WUSB54GP as a device it supports), it proceeds to uninstall the drivers. The workaround is to run the installer, and then when it errors out about not being able to find the device that needs the drivers, you need to open Task Manager and kill setup.exe. After that, just use Device Manager to update the drivers for the wireless adapter by browsing to "C:\Program Files\IOGEAR\IOGEAR WGU513 Wireless\Res". During the driver setup, you will be provided with a wizard of about 2-3 pages for configuring the wireless adapter -- just leave everything on its defaults and click next until you're finished. The settings don't seem to affect the operation of the adapter outside of Windows' own wireless configuration UI, so it probably can be safely ignored.

I've posted updated drivers for the Aureal Vortex2 (AU8830 chipset) on Rapidshare:
http://rapidshare.com/files/260484882/Aureal_AU8830_for_Vista_and_7_by_Guy_Paddock.rar

Calling them "updated" is a bit of a misnomer; I thought that the ones I previously posted were a later version than the ones that shipped with Windows XP, but I found out that post-SP2 for XP, some parts of the AU8830 drivers were actually of a later version. Also, I found that the last drivers I posted had a nasty paged pool memory leak that would slowly cause them to sap away the system's precious kernel paged pool memory, eventually leading the Windows session to become unusable.

I'm currently using these drivers in Windows 7 32-bit RC1, and they seem to work. They should work in Windows Vista 32-bit as well.

What I did for this new package was update all of the files with the XP (and, where possible, Windows 7) versions. The INF files have not been changed, but you should notice a more stable driver without the aforementioned memory leak. I have also stripped out the setup program for A3D and the A3D SDK as the INFs never referenced them, and they added about 10 - 15 MB to the download.

WARNING: On my system, with these drivers installed, if I do a "Scan for hardware changes" in Device Manager, my system goes to a blue screen, reports "NMI: Parity Error", and halts. If you encounter this problem as well, please comment on this post. My hardware is already a little odd as it is, so I want to know if it's the driver itself that causes this (in which case, other people would see the same behavior), or if it's just my hardware causing the problem.

NOTE: These drivers still do not have A3D enabled. A3D did not work with the XP drivers, and I honestly just care that I am still able to use this card in Vista and Windows 7. Please do not contact me about getting A3D support enabled, as I don't know enough about it to help.

ALSO NOTE: These drivers are 32-bit only! They will NOT work under Vista 64-bit or Windows 7 64-bit. The only way for me to add 64-bit support would be if I had the source code to these drivers and could re-compile them for 64-bit addressing. However, I don't have the source code, so I can't. Please don't contact me about getting 64-bit versions of this driver, as they don't exist for Windows, AFAIK.

Eclipse is my IDE of choice, especially for Java development, but one of the issues I've always had with it is that no matter what machine I use, it always seems to run dog slow, especially compared to Visual Studio. As soon as you switch from Eclipse back to VS, you see what I'm talking about – VS just seems to glide along effortlessly for most common tasks (with the exception of changing project settings or building).

Well, if you're like me and you have a workstation with > 2 GB of memory to spare, you can tweak Eclipse's Performance by increasing the maximum heap size to 768 MB, and the "PermSize", which CauchoWiki describes as "the maximum size for the permanent generation heap, a heap that holds objects such as classes and methods", to 512 MB. To do this, right-click on the shortcut you use to start Eclipse (usually a Desktop shortcut or Start Menu shortcut) and add the following parameters:

  -vmargs -Xmx768M -XX:MaxPermSize=512M

With these two settings applied, you should notice an immediate improvement. Files and windows will open faster, files will scroll more responsively, and auto-complete will open a tad faster. On the other hand, if you only increase one setting without the other, you won't really see that much of an improvement.

As always, YMMV, especially if you have a lot of other programs open, or have a ton of Firefox tabs open at the same time, as Eclipse will starting paging.

If you download Windows Server 2008 from Microsoft Academic Alliance, you may find yourself scratching your head when Product Activation fails after installation, with a cryptic error like 0x800705B4 - "This operation returned because the timeout period expired."

The answer is in this entry of Jim Blizzard's blog:
http://snowstormlife.com/blogs/bliz/archive/2008/02/11/activate-windows-server-2008-rtm-msdn-subscriber-downloads-version.aspx

Apparently, by having Windows attempt to change the key, and then providing the exact same key as provided during installation will cause activation to succeed.

My guess on why this is necessary: when you boot into the Windows Installer, there is no network connectivity, so setup cannot check for a local activation server, but since it sees that it's a Volume License Key, it tells Windows that activation will be processed locally after installation. Then, after install, Windows searches for a local server to no avail and eventually times-out. By re-entering the key after install, Windows can identify the key as an MSDN Subscriber / MSDNAA key and activate with Microsoft directly. But, this could just be wild conjecture.

Either way, I agree with Jim Blizzard -- this is totally counter-intuitive.

UPDATE (7/27/2009): I just realized that I have the WUSB54GP, not the WUSB54G. I have updated this entry accordingly, as the two are apparently different devices with different chipsets and drivers (thanks Linksys for that confusion...). These instructions are meant for the WUSB54GP.


In late 2004, I purchased a Linksys WUSB54GP (v.1) Portable 802.11b/g 54-Mbps USB Wireless adapter from CompUSA. Although it has a nifty little USB extension base that allows you to position the adapter wherever the signal is the strongest, the device itself has never performed that well. No matter what USB controller, operating system, and router I've used it with, I've always observed that the Linksys drivers just plain suck -- on high bandwidth downloads, it seemed like the device would actually get stuck, requiring me to disable and re-enable it to get it working again.

I also noticed that for the past two months or so when I was running Ubuntu Intrepid Ibex on my desktop system, my wireless was significantly more reliable (random hard lock-ups aside -- writing drivers for SMP is hard!). The WUSB54G uses the ISL 3890 chipset, which is supported by the Prism 54 project. In addition to more reliable connections, I noticed that with the Linux drivers the activity LED on the device does not light up as brightly, and it flashes at different intervals. The reason? The project must be using different firmware from what Linksys uses in their drivers, and that firmware doesn't seem to suck as much.

Sadly, at the beginning of the week I switched back to using Windows XP. I actually haven't used XP on a personal machine in about a year or so, as I've been using a combination of Vista and Linux. I had forgotten how simple and reliable wireless is in XP, compared to the overhaul that Vista really needs in the wireless networking area. Vista's wireless doesn't work half the time, and the fact that the "Network Diagnostics" feature always thinks it knows better than you makes fixing problems nearly impossible.

XPs better wireless support aside, I was back to the standard crappy Linksys drivers that drop the connection with high-bandwidth usage. My success with Linux got me thinking, though... If better firmware is really all I needed, and the chipset inside the device was likely used for many other USB wireless adapters from other manufacturers, perhaps I could just swap out the Linksys drivers for another manufacturer's more stable ones. Sure enough, that's what I did.

The Prism54 project noted that one of the other wireless adapters using the chipset is the "AirPlus G DWL-G122" from D-Link. After having several very stable D-Link products, I trust them, so I decided to give their drivers a try. Out of the four revisions of the DWL-G122, I downloaded the drivers for revision A (seriously, why do manufacturers re-use model numbers so much that consumers have to keep track of revisions?). I noticed that XP is more finicky than Vista about accepting drivers that do not have a matching vendor ID and/or device ID, so I had to change the device IDs in the driver INF to get it installed.

Fingers crossed, I told Windows XP to use the D-Link drivers and Voila! My wireless is now incredibly stable and I can use my full bandwidth without instantly losing my connection. I still have to occasionally reset the device, but only about once a day at most, compared to the 5-10 times per hour with the Linksys drivers (depending upon network activity).

Got a WUSB54G at home that's giving you trouble? Here's the review of the steps necessary to switch to the more stable D-Link drivers:

1. Download the drivers for revision A of the DWL-G122 from D-Link's website.


2. Extract the driver archive somewhere easily accessible, like a folder on your desktop. Do not run the setup.exe included in the archive!


3. Open up PRISMA02.inf and find the section entitled "[DeviceList]".


4. Change the text that reads "USB\VID_2001&PID_3703" to "USB\VID_5041&PID_2235" (the vendor ID and device ID for the WUSB54G).


5. Open up Device Manager and find the "Linksys Wireless-G Portable USB Adapter" device under "Network Adapters".


6. Right-click the device and select "Update Driver..."


7. Select "Install from a list or specific location (Advanced)" and click "Next >".


8. Select "Don't search. I will choose the driver to install." and click "Next >".


9. Click "Have Disk...", click "Browse...", and select the folder where you unpacked the driver archive earlier.


10. Select "D-Link AirPlus G DWL-G122 Wireless USB Adapter" from the list and click "Next >".


11. Windows will warn you that the drivers are not digitally signed because the INF file was modified. Click "Continue Anyway".


12. After installation, Windows may require you to restart your machine, especially if your wireless adapter was in use during this process. At your earliest opportunity, allow Windows to reboot so that the new drivers kick in.


13. You will likely have to re-connect and re-authenticate with any wireless networks you had previously connected to. Your wireless should now be much more stable. Enjoy!

Okay, admittedly, the phrase "Intrepid Ibex on iBook G4" has way too many "i"'s in it. Anyway, for anyone trying to get Intrepid Ibex on the system, you'll probably find that the installer can't find the install CD at the "Detect and Mount CD-ROM" step.

It took me about 15 - 20 minutes of searching this up, but I remembered previously having this problem when I had tried to install Drapper. The solution is to drop to a console prompt from the installer and run "modprobe ide-scsi". You should hear the CD-ROM drive spin up, and then you can exit the console back to the installer and re-run the CD detection step again.

Just an FYI for anyone also experiencing this issue.

EDIT: After installation, if you get a black screen, it's caused by usplash, the Ubuntu graphical splash screen. To fix it:
1. Reboot your machine (CTRL+COMMAND+POWER).
2. At the yaboot prompt, type "Linux nosplash" to boot your default yaboot configuration without usplash.
3. Edit yaboot.conf, adding:

xres=1024
yres=768

4. Run "sudo update-initramfs -u all"
5. Reboot and enjoy.

So, it's official, my parents are now addicted to BSG. We just finished the first season, and my mom says she has to buy the 2nd ASAP. I told her she'd like it if she didn't fall asleep during the miniseries, but she always acted like the show was way too hard to follow. Now, she's hooked, and the wild theories about how the humans will survive with Adama killed, and why he was shot, have begun.

Ah, BSG newbs. It's fun to watch.

Would you rather know everything, but have the power to change nothing; or know nothing, but have the power to change everything?

Ugh. Credit bureaus are such a fucking waste of time, resources, and money. They're greedy bastard organizations that get rich off of every single one of us by holding our (frequently incorrect) information hostage, and I think that our country needs to just outlaw them.

I've been helping my mom pull-up her credit report today because she was turned down for re-financing her current mortgage. Yeah, that's right -- the lender already gave her the mortgage several years ago, but they're un-willing to lower the interest rate because they're concerned with her ability to pay. That makes no sense! If she has to pay less each month, there's less of a chance she'll foreclose.

Dumb lender aside, one reason they say she was turned down was because of a $917 charged-off delinquent dept from American Express. There are three problems with that:

1. My mom hasn't had an AMEX card since 2004


2. The card she had with AMEX was settled, not delinquent.


3. The amount she owed AMEX at the time was about $1,500, not $917.

An additional problem was that she is listed as the co-signer for my dad's mortgage on our property at Keuka, even though my mom's name doesn't appear anywhere on the mortgage agreement and she wasn't even present when he took the mortgage out.

They said they acquired her credit report from TransUnion, so off we went to AnnualCreditReport.com (ACR), the website the big three credit reporting agencies created to supposedly satisfy the Fair Credit Reporting Act. We entered all of my mother's information, and asked for reports from all three agencies. Of the three, here are the results we got:

• Experian provided her with her report without any problem, but did not show any delinquent charges from American Express.


• Equifax refused to provide the report online as a result of a "problem authorizing the account for online access", and told her she would have to send them a letter by mail to receive her credit report in several weeks.


• TransUnion told her that she had an existing account with them and would have to provide the credentials for that account before she could access it, despite having reached the page from ACR which is supposed to be a back-door into the site. When we followed the link to request credentials, the site said a temporary failure had occurred that prevented us from obtaining that information. Their website provided no address or phone number, and the live support option was broken.

So, out of the three, the one we really wanted information from wasn't even functional. And, of the three, only one actually provided her with her legally required copy. It's good to see the Fair Credit Reporting Act is working. </sarcasm>

Beyond the fact that these sites weren't properly handling the session from ACR, I noticed the following issues with the sites:

• All three of the agencies require you to re-enter your SSN when you arrive at their site from ACR. If that's the case, what's the point of entering your SSN on ACR in the first place?


• Equifax does provide you with a free copy of your credit report if you go through their website instead of ACR, even though the whole point of creating ACR was that none of the agencies properly handled the annual report through their own websites.


• Equifax requires you to provide your SSN when navigating between sections of their website, and requires you to log-in to view your report after you just ordered it. They say they do this for security, but if someone stole your information, couldn't they just enter it again anyway? And, if you were stupid and walked away from your session, wouldn't a simple session timer be sufficient? You know, like the one my bank uses to protect my financial information, which I'd think is more important than my credit report.


• After coming from ACR, Equifax does not inform you if they detect that you have an existing account with them. Subsequently, when you order your report, you are asked to provide the credentials for the account you just created, even though you didn't just create an account, and asking to reset your password reveals your existing username.


• After logging-in, Equifax drops you at a report summary screen that requires you to click on each category of accounts you want to view. If you click "View Full Report", it sends you to the same page with a different layout. If you click "Print this report", you are sent to a printer-friendly page that provides only brief summary information about each account, with links under each to view full details. The links, however, go nowhere -- apparently, "printer-friendly version" means "version with unhelpful info and dead links".


• TransUnion doesn't provide you with their phone number or mailing address on the "Contact Us" page, but does provide you with a link to their knowledge base which has no information about how to reset your username or password in the event that their system is fucked-up and won't let you reset it (which, from my experience, is 95% of the time). Curiously, if you click on the contact information link for TransUnion from the ACR website, that same "Contact Us" page does provide the phone number and mailing address because it passes "phone=true" to the page. For a company that deals with information that can affect your future, that has to be the sleaziest, most under-handed thing for them to do.

We also found out that according to Equifax, that American Express debt of $917 that TransUnion reported about my mom shows up as a $917 dept from Bank of America. It's still not her debt, but impressive that both have different information for the same incorrect item.

According to the Federal Trade Commission, ACR is the only place you can go to get your yearly free report, so why aren't the tactics these unscrupulous, greedy companies are pulling illegal? How can we trust these companies when they can end up with this ghost information?

This has to be my favorite Stewie moment (wait for the end of the clip):
http://www.hulu.com/watch/35059/family-guy-quahog-mens-club

A Ruby developer who says Java's static type checking makes development more difficult is like a college student saying that the grammar check in Microsoft Word restricts his creative expression to just valid English sentences.

In fields where nothing grew but weeds,
I found a flower at my feet,
bending there in my direction.
I wrapped a hand around its stem
and pulled until the roots gave in,
finding there what I've been missing.
And I know....

So I tell myself, I tell myself, it's wrong.
There's a point we pass from which we can't return.
I felt the cold rain of the coming storm...

All because of you,
I haven't slept in so long.
When I do I dream
of drowning in the ocean;
longing for the shore
where I can lay my head down.
I'll follow your voice;
all you have to do is
shout it out!

Inside my hands these petals browned;
dried up falling to the ground,
but it was already too late now.
I pushed my fingers through the earth,
returned this flower to the dirt;
so it could live, I walked away now.
But I know...

Not a day goes by when I don't feel this burn.
There's a point we pass from which we can't return.
I felt the cold rain of the coming storm...

All because of you,
I haven't slept in so long.
When I do I dream
of drowning in the ocean;
longing for the shore
where I can lay my head down.
I'll follow your voice;
all you have to do is
shout it out!

All because of you.
All because of you.

All because of you,
I haven't slept in so long.
When I do I dream
of drowning in the ocean;
longing for the shore
where I can lay my head down.
Inside these arms of yours.

All because of you
I believe in angels.
Not the kind with wings,
no, not the kind with halos;
the kind that bring you home
when home becomes a strange place.
I'll follow your voice;
all you have to do is
shout it out!

-- "The Good Left Undone" by Rise Against

I recently found myself in the position where I had a Turtle Beach Montego II sound card with an Aureal Vortex 2 AU8830-based chipset that I wanted to get working in Windows Vista. Sadly, Aureal went bankrupt in 2000 and was later bought by Creative, so Aureal chipsets have had pretty poor support since XP -- the stock Vortex2 driver included in Windows XP was a bare bones implementation at best. One would think that Vista support of this ancient audio chipset would be a pipe dream, no?

Well, I knew that Vista supported using XP sound drivers. So, I found the last know drivers for XP -- the 5.12.2568.0 beta drivers -- and tried the installation. Vista choked on some missing sections and incorrect information in the driver INF files, so I modified the INFs a bit to suit the OS.

The result is here:
http://www.2shared.com/file/4322605/fd1d7a4c/Aureal_AU8830_25681__Vista__by_Guy_Paddock.html

Aureal Vortex2 8830 users, rejoice! You now have basic sound support in Vista.


UPDATE (7/27/2009): More stable drivers have been posted; please see my other post.


UPDATE (6/12/2009): MegaUpload is officially on my shit list for getting my original file confused with what appears to be some game. The download went from being about 18 MB to 459 MB. I've removed the MegaUpload mirror link.

UPDATE: I've been seeing some strange issues that appear to be related to this driver. Specifically, the Windows paged pool on my system seems to be growing slowly over the course of several days from about 150 MB up to 1.5 GB, causing Windows to eventually stop responding. The Windows driver verifier reported that this driver is attempting to free memory it has already released, which is causing problems for the kernel memory allocator.

This may be because these drivers were BETA drivers for Windows XP made by some of the developers from Aureal, and I would like to switch over to using the earlier official version of the drivers that Microsoft bundled with XP as it should be more reliable. Stay tuned.

So, I just had a 40 minute phone interview (if you could call it that) with a Microsoft recruiter, during which I was asked to write two functions in my language of choice and was asked only one other question (describe a project in which you feel you fully demonstrated your technical skill). As if the lack of an opportunity to "sell" myself wasn't disconcerting enough, I had issues writing both functions.

The first function was supposed to find the missing integer in a sorted array containing the integers 0..N. My first version just naively iterated over the array until it found the first number that was not equal to the index, and returned the index as the missing number. Then, the guy told me that I needed to make the function perform better. I asked him if the problem was with larger arrays, since it'd be iterating over more elements and would thus take longer, and he said yes. I then said I would want to do more research on the problem, but that I believed there was an algorithm, possibly involving a tree, in which one could recursively divide the array into two chunks and search each chunk (aka "binary search", but the name didn't jump into my mind). He told me to write the algorithm. I told him I needed to consult references, and he said he wanted me to try to do it without consulting references. I tried, and failed, to write it out directly. He stopped me and said, "Yeah, you would use a binary search, which is really what I was looking for here." Ouch!

The second function involved me replacing all spaces in a string with "%20" (i.e. URL-escaping spaces). He told me I had a character array, but that it was in any language of my choosing. I started to write the code to go character-by-character through the string until it found a space, and then I needed to allocate a new character array for all the parts leading up to the space + the two new characters + the last part. He stopped me and said, "instead of coding, can you tell me what steps you're following?" I told him I needed to find the space, remove it, and insert the %20 in its place. He said thank you, and then proceeded to questions. Ouch?!

I asked him how often on the Microsoft Office team he finds himself having to write or re-write low-level API functions like these, and he said there are many times when they have to carefully re-work Office API code to accommodate the functionality for new languages. He told me we had 1 minute left and asked if there were any more questions. I said "no", and he told me they'd be in touch.

That was the most painful phone interview I've ever had.

Why do I ever fucking touch hardware, when all I seem to do is break something?

GRAR! And why is what's broken so difficult to fix?