Note: This fix involves changing the Registry. If you’re not comfortable with that, you may have to just put up with wonky search results. As always, you should backup your Registry before making any changes, and I take no responsibility for any damage these instructions may cause.

I should also mention that I’m doing this on Windows 7 Ultimate x64. Other versions might require a different fix.

As I said, the problem is that the OneNote Search Connector DLL isn’t registered properly. On Windows x64, Explorer needs to know to look for the 64-bit version of the DLL, found in the “Program Files” folder (not “Program Files (x86)”). On my machine, the appropriate keys were simply missing from the Registry. These are the keys you’ll need to create:

Windows Registry Editor Version 5.00

[HKEY_CLASSES_ROOT\CLSID\{0875DCB6-C686-4243-9432-ADCCF0B9F2D7}]
@="Microsoft OneNote Namespace Extension for Windows Desktop Search"

[HKEY_CLASSES_ROOT\CLSID\{0875DCB6-C686-4243-9432-ADCCF0B9F2D7}\InprocServer32]
@="C:\\Program Files\\Microsoft Office\\Office14\\ONFILTER.DLL"
"ThreadingModel"="Both"

[HKEY_CLASSES_ROOT\CLSID\{0875DCB6-C686-4243-9432-ADCCF0B9F2D7}\ProgID]
@="OneIndex.ShellFolder.1"

[HKEY_CLASSES_ROOT\CLSID\{0875DCB6-C686-4243-9432-ADCCF0B9F2D7}\ShellFolder]
"Attributes"=dword:20180000
"WANTSFORPARSING"=""

[HKEY_CLASSES_ROOT\CLSID\{0875DCB6-C686-4243-9432-ADCCF0B9F2D7}\VersionIndependentProgID]
@="OneIndex.ShellFolder"

To make your life easier, you can download the appropriate registry file here. Just save the file to your desktop and double-click it to create the appropriate Registry keys.

Note: If your copy of Office 2010 Beta is installed on a drive other than C:, you’ll need to edit the Registry file to reference the correct drive before it will work.

Once you create the Registry keys, clicking on OneNote search results should take you to the appropriate page in OneNote. You should not need to reboot or rebuild your search index.

On my machine, however, there was one more problem. The “heading” under which the search results appeared in the Start menu was still labeled with a long, meaningless codename, prefixed with “oneindex14–”. It turns out the fix for this is easy, too.

Just open the start menu, type “%UserProfile%/Searches” and hit Return. In this folder, you should see a couple of items under the heading “Search Connector.” Find the one that’s named after the weird code, and simply rename it to “Microsoft OneNote,” like you would rename any other file. Your OneNote search results should now appear in the Start menu under that name.

Thanks to John Guin at the OneNote Testing team at Microsoft, whose original blog post about Office 2007 led me to cook up this solution.

But the most recent insult was the sudden, unexpected change of the User page. Previously, if you clicked on your username in the upper lefthand corner, it would take you to a page that listed your recent comments. Now you get dumped to a Firehose page, forcing you to do an extra page load to get to where you’re going — and don’t get me started on Firehose.

So I decided to write a Greasemonkey script to fix it. You can download the script here.

Basically, what the script does is rewrite the link on your username so that it takes you to the Comments page instead of the Firehose page, like it did before. Sorry if the code isn’t pretty — I’m really no expert on writing Greasemonkey scripts. If you have problems or suggestions, please let me know (post them below, for example).

Incidentally, if you have no idea what I’m talking about, investigate the Greasemonkey plug-in for Firefox. It allows you to run arbitrary JavaScript code whenever a specific page loads, which allows you to process and/or modify the content of the page in various ways.

And by the way… you did do a “View Source” to make sure that there wasn’t any malicious code in the Greasemonkey script you downloaded from someone’s random blog… right?

It turns out that the SFPL’s advanced search application will return false positives for nearby ISBN numbers if the one you’re searching for is not available in the system. I’ve now modified the script to do a more reliable two-phase search, where it only checks the advanced results if it detects a positive hit using the basic search first.

The script will also now take you to the more user-friendly basic search results page every time, even though it still uses the advanced search to check all of the branch libraries.

I’ve removed the XISBN functionality, which was designed to check other, “related” ISBN numbers if the one you’re looking at wasn’t found. This was a good idea in theory, but it seemed to return too many inappropriate results: If I’m searching for a book, I’m not necessarily interested in the large-print edition, or the audiobook, etc. Searching for these things just caused a lot of inappropriate and unnecessary queries. What removal of this function means, however, is that you’ll need to check the hardcover and paperback editions separately (but that often seemed to be the case before, anyway).

One feature I added was the ability to spot new books that are on order at the library but have yet to actually arrive. This can give you the chance to click over to the library and reserve your copy, even though the book isn’t actually in stock yet.

The other positive change is that I went through and did a thorough housekeeping on the code, which I probably should have done in the first place. I’ve eliminated the unorthodox and inconsitent indentation and line breaks and generally cleaned up the source, which should make it a lot easier to maintain in the future.

Let me know what you think!

This particular Pearl, however, was a replacement for my old Pearl — an earlier model 8100 — which I accidentally dropped in the toilet about two weeks ago. And that gave me an idea: Why not swap out the broken screen in the new Pearl with the working screen from the old one?

This was, of course, a crazy idea. Mobile phones are modern marvels of micro-scale engineering. They’re just not meant to be serviced by the public. If you even try to open one up, you’ll probably end up breaking something important and you’ll be out a couple hundred bucks. A plan that calls for taking two of them apart is doubly crazy.

So I charged on ahead.

The first thing I did was scour the Web for advice. Sure enough, I found a very informative video showing how to take apart your BlackBerry Pearl. Next, I hopped down to my local hardware store and bought a Torx T-6 screwdriver, which is apparently the size you need to open all kinds of mobile phones.

I’ll spare you the blow-by-blow of the process of disassembling the phone, because the video explains it nicely. I can tell you what it was like to actually do it, however: nerve-wracking. I reiterate my statements from three paragraphs ago: You’d be crazy to try to do this. I don’t recommend it. In fact, I recommend against it. (And bear in mind that it voids your warranty — but we’ve already established that your warranty is pretty useless if your LCD is already broken.)

Note, however, that if you don’t happen to have a spare Pearl lying around, you could perform the same operation with a replacement screen that you order online. They cost about $40.

Some observations:

First, taking off the bottom piece of plastic is the hardest part. It’s not immediately clear how to do it, and once you start messing around with it you realize quickly that it seems pretty easy to break. In the video they use a small screwdriver to pry off the tabs. I’d try to avoid this. In the battle between metal and plastic, plastic inevitably loses. In the end I felt the most confident just prying it off with my thumbnails. The tabs that hold it on aren’t really locked into anything. You just want to use the natural pliability of the plastic to pop it off — carefully.

You should use caution when removing the side plastic pieces, also. Go gently. I found that when you get to a certain point  they will suddenly fly off in a way that’s a little startling, but if you’ve done it right they won’t have broken.

Now you get to the nerve-wracking part, when you realize that you’ve just taken apart your phone and you’re about to fiddle around with the screen and the tiny motherboard. Don’t get spooked — there’s no point in turning back now.

In the video they point out a small piece of tape over the ribbon connector for the screen that needs to be removed. It was there on my 8120, but oddly enough the 8100 had no piece of tape. Because of this, I didn’t really worry about saving the tape from the 8120. If it wasn’t installed in one of the phones then it must not really be necessary.

Try not to touch the LCD. Pretend that touching it will make it shrivel up and die. But in reality you will get a few fingerprints on it. For that reason, you’ll want some kind of lint-free cloth available to wipe it clean before you put the phone back together. I used a chamois that came with my Asus Eee PC 901.

Removing the screen is the most nerve-wracking part of all. It’s actually stuck to the motherboard with a big piece of double-sided tape. You need to loosen that tape slowly, very gently, so as not to damage anything. For what it’s worth, the tape seemed a lot stronger on the 8100 than the 8120 — perhaps the adhesive gets tougher with age.

Here’s where your little flat-head screwdriver comes in handy. Get it underneath one of the corners of the LCD and pry ever so gently. Work your way around the screen, doing your best to touch as little of the motherboard as possible. Don’t dig deep underneath; just work around the sides. If you’re doing it right, you should hear a sound almost exactly like transparent tape coming off the roll.

Once you’ve got the screen off, if you’ve remembered to flip open the little latch on the ribbon connector, it should slide right out. You shouldn’t have to touch the ribbon cable at all.

If you’re planning to re-use a screen that you’ve removed from a phone, remember to lay it glass side down, because the adhesive will stick it to your workbench. Try to get a minimum of lint and crumbs onto the tape, also.

Replacing the LCD isn’t difficult. The ribbon cable should slide right in with no pressure required. Remember to close the little black latch to lock it in place. Hold the LCD by the sides and give it a little squeeze against the motherboard so that the adhesive sticks. It’s at this point, also, that you’ll want to clean any fingerprints off the LCD.

The case reassembles in exactly the opposite order of steps that you just took to disassemble it (naturally). The side pieces seemed a little fussy; squeeze them as you re-insert the Torx screws to make sure everything is in place. Everything else pretty much snaps together. If you’ve forgotten which end is up on some of the pieces, the ports and connectors should make it obvious.

Now, if you’ve done everything right, you can replace the SIM card and the battery and you should have a working phone with an unbroken screen. I know I do! Total cost of repairs: $4.50 for the Torx screwdriver.

Allow me to reiterate once more, however: Unless you’re prepared to spend $200 on a new phone, don’t do this.

For the truly geeky among you: In the latest post to my Fatal Exception blog over at InfoWorld, I’m taking a look under the hood of Chrome, Google’s new Web browser. A lot of articles make mention of how Chrome is open source. I actually put it to the test, by building a custom copy of it myself.

Along the way I found out a lot of interesting information about Chrome’s internals and how Google built it. On the plus side, it’s very clean, well-organized code. On the minus side, it looks like it’s going to be Windows-only for a good while, yet.

Anyway, I had a lot of fun doing this piece (it’s been a long time since I had a legitimate excuse to pull out a compiler on Windows), so if you have the hobbyist spirit, drop on over and join the discussion.

As usual, you can download the script here. You should delete the old version by hand before installing this one, because I’ve updated the namespace to reflect my own site instead of Jon Udell’s (finally!).

If you don’t understand what this is all about, check my earlier post on the subject. (And if you don’t live in San Francisco it probably won’t be much use to you anyway. Try a Google search for a similar script that works with libraries in your area.)

If you have any questions or problems with the script, post here and I’ll try to address them.

Notes: These torrent files do not contain the OS itself. You will need a BitTorrent client to download the actual disc images, and you may also need blank CD-R media to burn them.

If you have a regular Pentium-type system, you should probably choose the i386 option. Only choose amd64 if you know you have either an AMD or Intel 64-bit chip (and you’re sure you want to use the 64-bit version of the OS).

Most people will probably want the desktop version; if you don’t intend to use Ubuntu primarily to run a Web server or some other “back office” purpose, you don’t need to download the server disc at all. You can still install a personal Web server (or any other software) on the desktop version.

(UPDATE 1 May, 2008) – Now that the initial rush has quieted down, I’ve replaced the links listed here with links back to the torrents on the official Ubuntu site. You can see still further download options on the Ubuntu 8.04 release page.)

Here are the links:

• Ubuntu 8.04 Desktop/i386
• Ubuntu 8.04 Server/i386
• Ubuntu 8.04 Desktop/amd64
• Ubuntu 8.04 Server/amd64

If you want to know more about this script and its history, you can check my earlier post on the subject.

If you just want to dive in and start searching for library books via Amazon, you can download the latest script here. (Note that you must have the GreaseMonkey extension for Firefox installed for this to work.)

It occurred to me this afternoon that it might be nice to have a native Ubuntu Linux partition running on my main Windows Vista machine. “Should be easy,” I thought. So I whipped out my trusty Feisty Fawn CD, resized my Vista partition with GParted, and proceeded to install Linux.

Big mistake.

The installation itself went fine and Linux started right up. The problem came when I tried to boot back over into Windows Vista. In a nutshell, it wouldn’t. I’d get the green “Knight Rider” screen, but then the system would switch to a black screen and everything seemed to grind to a halt.

The problem got worse from there. Supposedly there’s a Startup Repair feature on the Vista install DVD. Unfortunately, I was never able to look for it. Booting from the DVD only got me as far as the same black screen.

Some Web searches revealed that Vista apparently hates it when you muck with your partition tables using older tools. This goes not just for Linux, but Partition Magic and the like as well. Just about any tampering with the drive after Vista has been installed results in a system that cannot boot Vista.

(This is yet one more reason why many of us ask why on Earth we installed the stupid thing in the first place; but I digress.)

Luckily I wasn’t hosed yet. As I said, I had a working Ubuntu partition on my drive now. Fortunately there are tools under Linux that can fix this very problem.

The package you want to look for is called ntfsprogs. It’s not installed by default with Ubuntu, but it’s available from the official package repositories. You can download it using Synaptic or do a “sudo apt-get install ntfsprogs” from the command line. It’s available for most other distros as well, but I’ll leave it to you to figure out how to get it. (Hint: You can start with the project homepage at http://linux-ntfs.org.) The important thing is to make sure you have the latest version (1.13.1 as of this writing).

Once you have ntfsprogs installed, the rest should be easy. Assuming you know the name of the partition that’s having the problem (/dev/sda1 on my Ubuntu system), you simply open a terminal window, unmount the partition in question, and type “sudo ntfsfix partitionname“. It should fix the problems with the partition table in just a few seconds.

Once ntfsfix works its magic, reboot into Windows Vista. Windows should automatically force a chkdsk process. When that completes, you should be able to dual boot between Vista and Linux to your heart’s content.

Jon originally wrote the script for the libraries in his hometown of Keene, New Hampshire, but there are only so many different library catalog systems in use. The script is adaptable to any of them, and lots of folks have modified it to work with the systems in their own communities. I never found a good version for San Francisco, though, so I spent some time working with Jon’s latest code to get it working. It needed just a couple of tweaks.

You can click on the image below to see what an Amazon page might look like with the script installed:

To get the script working, you need to install the latest version of the Greasemonkey add-on for Firefox. This is the engine that runs the script automatically whenever you load an Amazon page. With Greasemonkey installed, you should be able to click the following link to automatically install the LibraryLookup script for San Francisco public libraries. It should begin working immediately.

The script isn’t perfect. It doesn’t sort the due dates on checked-out copies of books, for example, so you’re not always given the earliest date that a copy of a book will be back on a shelf. The SFPL’s system also doesn’t display every branch that has a copy of the book, only the ones it deems “most relevant” (which seem to be more alphabetical than anything else). And it can occasionally be a little sluggish. But, in general, this system gives you a good idea as to whether a given book is in the library system and is readily available for check-out. Take it for a spin and let me know if it’s useful to you.

UPDATE (2 April 2008): Version 1.2 uses a more complex query, which should return results for all of the branch libaries (instead of just the first few).

UPDATE (1 May 2008): Version 1.3 should now work with Firefox versions 2 and 3.

*The latest is that Jon claims to have gotten his script working with Internet Explorer, instead of just Firefox, but I can’t vouch for that. Read his blog post on the subject and see if you can get it working, if IE is your browser of choice.