[Solved itself] Windows cursor is broken.

[Bobbias]

Ok, this is a big deal. I don't know how, or what caused it, but my cursor is fucked. I've already tried changing it, and it does "change" but not in any useful way.

As far as I can tell, it looks like instead of drawing the cursor the right way, it's only drawing the top 3 or 4 pixels worth, and then tiling that downward. I can't really screenshot it, since the cursor never shows up in screenshots, but when it's not over text, it's extremely difficult to see.

I noticed the change when I used the touch pad to bring up my desktop after I'd let the laptop sit long enough to turn the screen off (I have it set to 5 minutes). It was fine before, but the moment the screen came back on my cursor was fucked.

EDIT: The problem seems to have solved itself somehow. I downloaded TrendMicro's Housecall, started it, ran HijackThis, closed IE (which I had opened to get Housecall, because I forgot it's not an in-browser thing any more) and apparently closing IE fixed it. (I hadn't even told HJT to delete anything yet, and Housecall was still just starting to scan.) This is easily the weirdest issue I've ever had when it comes to computers.

[Spectere]

...weird.

I've seen that happen when video card drivers have a minor freak out.  Really odd that closing IE, of all things, would have fixed it.

Are you using IE9 beta or IE8?  IE9 will use your graphics card for rendering pages, so if its renderer flipped out I suppose that could have caused something like that to happen.

[Bobbias]

It was IE8, and the rendering glitch happened before I ever opened IE. All I had open when it happened was Chrome and Notepad++ (well, and Winamp and random background processes).

But yeah, that has to be the weirdest glitch I've come across so far.

I should point out that I had disconnected my mouse a couple minutes before then screen turned off, and hadn't touched the touch pad till I woke it up, and the glitch showed up. That could have had something to do wit it, but that is still a weird glitch.

And when I changed the cursor, the glitch still applied to the new cursor, so you could see tiny variations in what was being drawn, but the actual drawing glitch was still there. If I switched it to the animated cursor pack, you could see the multicolor animation thing going on on all the little cursor pieces it was rendering (for the few pixels that were visible).

It also did weird things when it came close to the top of the screen. The closer I got to the top (within maybe 10 or 20 pixels), the lower the area of the pointer that was drawn would get. I noticed that when I came close to the top of the screen, the cursor would change into the "working" cursor, and as I moved closer to the top of the screen, the rendered section would be lower and lower down the circle. And when hovering over text, it would repeat the top section of the text cursor (which was a bit more visible).

In any case, it seems to be gone now.

And Housecall found a generic trojan, and a spyware, so I guess it was a good idea to do the scan, lol.

[Spectere]

Definitely sounds like a video card driver freakout, then.  I think Windows draws the cursor in an overlay, so if there was an overlay bug in your driver then that could have done it.  Maybe closing IE didn't necessarily solve the problem, but for whatever reason triggered the video card driver to reset the overlay.

[Bobbias]

Maybe. It still weirds me out that something like that would randomly happen. This laptop has been working perfectly (aside from the usual slowdown issues accompanied by running a computer for any length of time) since I bought it, so it was really bizarre to see something like that just pop up and disappear. I guess it could be some really obscure glitch in the video drivers. Haven't updated them in a while, so maybe I should.

[Spectere]

Yeah, it probably is.

I had something similar happen on my old Compaq laptop, running XP, so it's definitely something that can happen.  That said, I had that laptop for about two years (before getting my Dell, which I've had ever since <3) and that only happened once or twice.  I wouldn't worry unless you start seeing it happen regularly.

[Bobbias]

Yeah, I haven't seen it since then. Actually, this kinda reminded me of how little I know about the windows internals dealing with the cursor and mouse interaction... I really don't know jack all about that part of windows (not that I REALLY know much about too many other parts, but most other parts I have some idea of how things fit together).

[Spectere]

Yeah, I'm not entirely sure about any of that, either.  Mainly just a theory on my part, because I've seen it happen before.

Windows is a pretty fascinating beast, actually.  One of the things that I respect the most about it is how it manages to have so many compatibility shims (it literally has built-in compatibility profiles for about 8500 programs -- download the application compatibility toolkit to view them) and other workarounds for other vendors' shoddy code without any of those compatibility "fixes" impacting modern software.

I try to pick up as much as I can about its internals.  I used to enjoy reading Sysinternals (prior to it getting gobbled up by Microsoft and getting updated far less frequently) because there were so many neat things that they documented.

[Bobbias]

I've seen sysinternals before, and read the odd post, but never really followed it. Windows really is an interesting beast, and as much as I bitch about it sometimes, there really is a lot of good work that goes into it. Unfortunately, there's also stuff like COM, lol.

[Spectere]

COM was good when it was released.  When you consider when it was made, it was definitely workable.  Also, when it was becoming stale and annoying, it was effectively replaced.

A lot of people complain about things like the Windows API, but when you get right down to it, Microsoft has been steadily replacing all of those APIs with newer technologies.  That's why Windows manages to stay compatible with ancient software yet can also support shiny stuff like DirectX and .NET.  Free products like Visual Studio Express and its incredibly liberal licensing (you can see the Express editions with commercial products, believe it or not) only help promote these technologies.

The positive result of all this is that you can still run the aforementioned ancient software on your shiny new Windows 7 box.  This is something that you really can't do on anything else.  Mac OS held out for a very long time (I can run Mac software compiled for 68K on my 2002 PowerMac G4 with hardly an issue) but nowadays you need a third-party emulator to run anything older than PowerPC OS X software, and G5-specific software doesn't, and probably never will, work.  Linux can theoretically run anything created for Linux and, IIRC, SysV binaries, but if the binaries aren't statically linked and source isn't available, it's going to be a total nightmare getting the ancient libraries that you'd need to support such a thing.

If you want some insider information about Windows, including old application compatibility and just the reasons why some things are the way they are, check out Raymond Chen's blog, The Old New Thing.  He wrote a book, presumably with similar content that he has in his blog.  I'm thinking of purchasing it.  His blog is definitely an excellent read, though.  I check it daily (yay Google Reader).

[Bobbias]

I've been to that blog a couple times, actually. I always forgot to favorite it or anything, but I recognize that layout (and the name is familiar enough).

I've gotta agree that while there are times that microsoft has dropped the ball, overall they're doing much better than many companies do when it comes to maintaining APIs, documentation, etc. And Visual studio kicks the shit out of it's competitors. I don't see why anyone would use anything other than the VS suite.

[Spectere]

Visual Studio is my favorite IDE thus far, and I've used it (every version since version 6, which I thought was very, very lacking), XCode (quite nice; comes in a close second), KDevelop (seems okay, haven't used it too much), emacs (holy keyboard shortcuts, Batman), NetBeans (it's...okay), Eclipse (I really, really hate it), a few random PHP IDEs, and, uh...Notepad++ (mostly for QuakeC) and nano (mostly for when I'm feeling masochistic).

My favorite part about Visual Studio is that it generally gives you a decent amount of assistance in both form and API documentation without being a complete shithead about it like Eclipse and NetBeans can be (those two will add in closing braces and various other elements and you can't turn that behavior off in some languages).  XCode is actually a bit better when it comes to helping you fill out function calls, at least with Objective-C (basically, you type the function and you can tab between parameters), but it's a bit worse when you can't seem to remember every last character from the function name.  It doesn't have a nice scrollable box like the newer Visual Studios do.

Visual Studio is also great for its extensibility.  Not so much with the fairly trim Express editions, but the higher end editions let you snap in just about anything.  You can get extensions to add support for all sorts of version control systems, database systems, entire languages, you name it.  The fact that it works rather well with dual monitors is very nice as well.  Oh, did I mention that I love its debugger?  If not: I love its debugger.  It hasn't failed me yet.

XCode has come a long way since its original inception as Project Builder, though.  After I (mostly) got over its unusual interface I found myself starting to like it a little.  I haven't done a whole lot with it as of yet (I fiddled with a GLQuake OS X port back in the Project Builder days and am working on an iOS project now at work) but it's starting to grow on me now (so is Objective-C, which has really odd-looking syntax for a C-based language).

[Bobbias]

I've never really used the debugger in Visual Studio. I mean, I've used it once or twice on my own program to debug simple issues, but nothing beyond using it to watch a variable or two to check if they are working correctly. Anything more and I've used OllyDBG, which I quite enjoy. A little while ago I actually found the exact jump location in a program that I needed to bypass to bypass the serial checking function to play pirated copies of the game. It was actually kinda fun to do, but I am nowhere near capable to do much more than that in a debugger. Even that was quite difficult for me to figure out how to attack. (I eventually searched for the error text string that appears when you input a wrong code, found the function that calls the error box, traced back to the calling function there and replaced the necessary jump (well, patched a JNE to something that was equivalent to a JE, because it wouldn't let me replace it with a JE))

[Spectere]

I've used the Visual Studio debugger extensively in both C++ and the .NET languages.  It was a real life-saver when I made that old KKR thingie for StepMania.  StepMania 3.9's code was a bit on the convoluted side and had references to specific scoring systems in odd places, so the debugger helped me see every single area where the assertions would fail.

Similarly, it's been a godsend when I've had to find the source of exceptions in my C# applications.  I ended up writing a decent sized piece of software at work a few months back and found that the debugger was a great way of seeing where users could break my code (i.e. mash a few keys, hit ENTER, and see what breaks...I've learned that that's not far from what some people do).

The advantage of doing all of that is that it's source-level.  In C# you can even make certain changes and continue execution, letting you literally patch on-the-fly.

I've also used OllyDBG when I didn't have the source code available, but the only practical thing that I've used it for was patching some annoyances out of the O2LX O2Jam server.  That was kind of a fun project, and was pretty much justified since the program has an annoying banner that comes up in the game client for a long time that cannot be turned off (well, not normally, anyway).  The first issue was that the file was compressed.  Not only that, but the compression engine that it used had anti-debugger code in it.  Regardless, I was still able to extract the raw binary image with OllyDBG.  Second, I found the relevant code that displayed the banner and killed the jump with a couple of NOPs.  Lastly, the programmers put in a check to see if the program had been modified.  That was a bit tricky to find, but I able to JMP over the check after I located it.  I think that Milk Chan's O2Jam server used my modified O2LX binary when it was up.

I also have a lot of fun dissecting data files in a hex editor.  I wound up actually buying Hex Workshop when I was still in college because it worked so well for me.  I wound up documenting most of the data files for the insanely fun DOS game, JetPack, for example.  I suppose you could say it was a bit of an obsession for a while.   I also worked out the package format that Oregon Trail for DOS uses, allowing you to extract the game's animation scripts and graphics and make simple mods.  It also uses PCX files for its graphics, which I also wound up having to examine closely to get it to work because the official PCX specs kind of suck (I believe one part of them was downright wrong for 256 color PCX files).  The end result was a simple PCX file viewer.  Sadly, all of the in-game strings are hard-coded into the executable, but that doesn't stop you from changing the diseases to STDs if you poke around a bit.  I should really jump back on that project and work out their animation format...

[Bobbias]

You don't happen to still have O2LX/O2SX do you? I haven't been able to track down a copy in ages, and would love to have it. Originally Milk Chan gave me his copy of it, but after a while I ended up losing that.

Also, it's kinda funny, I actually talked with the creator of that, Vittee, for a while before he made it. I was on a forum that was trying to create a server by tracking all the packets and building a server from that, and he was a member of the forum. He'd made a couple claims that he could help, and told them he was willing to possibly work on something, and then disappeared for a bit, and then Kylecito announced his server, and shortly after Milk Chan got it. Vittee didn't really want it released at first, but eventually was OK with it, IIRC. Now he's actually working with one of the few O2jam servers as a developer for their custom server.

But yeah, I've been looking for that shit for a long time, and figured it was finally lost to the public side of the internet. I even went as far as to ask Turkyslam for it on the PH boards, since they took their server down, and he was given that server by Milk Chan.