UPEK Protector Suite strikes again

I thought (or perhaps naively hoped) UPEK had fixed all their issues. Alas, I was to be disappointed on a large scale as you can see below.


I admit, it's not a regular occurrence, but once is more than enough..

I'll try Authentec next time..

ITaskbarList3::ThumbBarUpdateButtons complaint again

I'm going to complain about it again (first time). I recently saw a crash log where the user had right-clicked on our application's notification area icon around the same time as my call to ThumbBarUpdateButtons - the result was a modal TrackPopupMenu call within the ThumbBarUpdateButtons call.. !!

What a complete mess - so make sure you only call ThumbBarUpdateButtons from a posted message/equivalent specifically for that purpose (at least in Windows 7 RTM). Unless you like strange (though rare) crashes, and the possibility thumb buttons being in the wrong state..

Windows 7 RTM feedback...

I decided to go ahead and buy a TechNet subscription so I could "evaluate" Windows 7, and am now running Windows 7 RTM on both of my computers. One was an upgrade from Windows Vista 64-bit, the other was a (sneaky) forced upgrade from Windows 7 RC.

In the spirit of "evaluating" Windows 7, I thought I might post some of the issues/problems I have encountered in the RTM version:

• The upgrade process reset the date modified attribute for all the directories in my user folder. That was one of my favourite things to sort by, so definitely an annoyance.
  
• The installer asked me to remove Avira AntiVirus which wasn't actually installed anymore. It was in fact complaining about some old left over driver which I had to manually remove - but how they expect the average user to work that out I don't know.
• I had a printer shared on the Windows Vista installation. I had renamed the share (to change a space to an underscore, because I was trying to make it accessible from another Mac OS X computer). After upgrading from Vista to 7, the share was still there but it didn't actually work anymore. In Devices and Printers, the printer wasn't marked as shared so I had to share it again. Unsurprisingly, I couldn't use the old share name. I am now left with two printer shares, one of which doesn't work. I thought I would try and remove the dead share using the "net share" CLI command, but I just get: "System error 1801 has occurred. The printer name is invalid." Lovely.
  
• This one was on the machine I did the naughty upgrade from Win 7 RC on, but I can't say if that was related to the problem or not. I had .NET Framework 4 beta 1 installed on Win 7 RC, after upgrading to Win 7 RTM I ended up with a broken .NET Framework installation. The solution here was to uninstall .NET Framework 4 and then reinstall it.
• When dragging an object from an external application (say foobar2000) Windows Explorer windows no longer automatically scrolls when you are near the edges of the items area.
• When dragging an item over an empty area of the task bar, it no longer shows the desktop automatically. This is probably due to the fact that you can now pin items to the task bar - my view though is that it could very well do both. You can now go over to the show desktop button, but that is all the way in the bottom right corner..
  
• This is definitely annoying, it happens on my desktop with a WUXGA monitor. Since it is reported on both AMD ATI and nVidia graphics cards, I'm reluctant to blame the graphics card driver (which is the latest version).
• When connecting say an SD Card, it doesn't prompt you to view the files or do something with it. I think they have changed this in Vista now also. There is probably some way to change this, I will have to have a look, but generally I want to look at the photos or something when I put in the card from my camera..
  
• I wrote about it before, but the removal of the columns header in the Windows Explorer views other than Details is a major step backwards. You now can't at a glance see the current sort order. Sorting by the reverse order of another criteria now takes a lot more work.
• I wrote about this before, but I still think the Show Desktop icon should be near the bottom left of the screen. The desktop icons are usually on the left of the screen, and also I like liked to blindly click in the bottom right to view the calendar.
• Amazingly, the Windows Photo Viewer doesn't handle smooth (Microsoft) mouse wheels correctly. It falls into the category "doesn't react unless you scroll the wheel exceptionally fast", and the bug I believe is a is a carry over from Windows Vista. My post on WM_MOUSEWHEEL failings is currently the fourth result on Google.com for "WM_MOUSEWHEEL", so hopefully I've made the world a better place regardless :p
  
• Some of the changes to Windows Update seem a step backwards to me. On the page where you select which updates to install, the published date column has been removed, so you have to click on each individual update to see the date. There is also now no "Install" button on that page, you are returned to the previous page to click on install which is a bit more mouse work.
  
• Somehow this UI glitch wasn't spotted:
  
  
  Yes it is minor enough (the "System" text overlaps the mouse over background), but surely you would notice that from the get-go?
• The new calculator also tends to slow me down. Vista essentially has the "Scientific" and "Programmer" combined in one, and as I use both views all the time this was really useful. They are separated in Windows 7 which means constant switching of the view, which also causes you to lose your current number (it did before, but I never needed to change view). If they had a toolbar with buttons to change the view, that might help. Animations that weren't quite as annoying after the tenth time would also help. As I mentioned previously, the swapping of the View and Edit menus is also a nuisance. (Some of the new parts of the calculator are useful though, for example the binary display in Programmer view).
• It really seems unnecessary/mean to remove the Vista Ultimate Extra games (Hold 'Em etc.) when upgrading to Windows 7. It should provide compatible versions.
  
• The best is saved until last: On the Windows Update restart reminder, the "Remind me in" time defaults to 10 minutes. If you click on the drop down (mouse left click), press down a few times to four hours (keyboard down arrow), click on the drop down again (mouse left click) and then click on Postpone (mouse left click) it will prompt you again in 10 minutes - not four hours. Not the first time mixed keyboard and mouse navigation has caused issues. This was probably an issue in Windows Vista too but I only recently worked it out. At least I now know why that prompt annoyed me to no end!

OK, I think I got most of it out of my system :p

Braindead logging

Fun with ITaskbarList3::ThumbBarUpdateButtons

This issue just cropped up:

-We are in function A (a callback function)
-We call ITaskbarList3::ThumbBarUpdateButtons
-It calls SendMessageTimeout with the SMTO_NORMAL flag and not SMTO_BLOCK
-Another instance of our process does a SendMessage to the first instance (for command line processing in this case)
-That results in function B being called. But its not legal to call function B within function A!!

We end up with a boom of kinds.

The end result is that I will have to make all calls to ITaskbarList3::ThumbBarUpdateButtons from a custom message posted by PostMessage - because in any other case it can open up holes I don't want (even if they are rare holes....)

Update: As far as I can see, Shell_NotifyIcon does use the SMTO_BLOCK flag. So why ThumbBarUpdateButtons doesn't I don't know..

Blurry text disease strikes again

A little while ago I wrote about how the Windows 7 RC jump lists are affected by such a problem. I decided to install Visual Studio 2010 Beta 1 and to my horror, it has also been struck by this illness in several places. Apparently something to do with that thing called WPF. Have a look at this over-sized blown up comparison of the main menu between VS2008 (top) and VS2010 Beta 1 (bottom):

Link to comparison screenshots

Look at that N, p and r! The Intellisense drop-downs are also quite bad.

How anyone can think that is remotely acceptable is beyond me, though apparently they are going to fix the Visual Studio problem for Beta 2.

Dan K kindly blew up my Windows 7 RC screenshot, and as you can see the problem is not nearly as bad there: http://shots.kordix.com/20090714_1.png. Still not great, though. Seems like they might have fixed the Windows 7 issue in builds after the RC, but the real test will be when the final version is out and installed here...

PSU whine/squeal/buzz... (under graphics card load)

Just wtf is this about. I'm referring to this, or this, or this, or this... It's horrible, and loud. Seemingly known as 'coil whine', but it's some combination of a whine, buzz and squeal.

I had it when I had a GeForce 9800 GTX graphics card - but it wasn't that loud and happened in fewer situations. At that time I never realised it was coming from my PSU, I just assumed it was my graphics card. Then I read about the PSU issue. Then my GeForce GTX 275 turned up and the noise was happening in far more situations and it was a lot louder. Then I double checked where it was coming from and indeed it is the PSU..

This was one of the more informative things I read about it. But I'm going to blame my PSU since it happened in some capacity with two fairly different video cards. After reading lots of PSU reviews (jonnyGURU.com at least made that slightly less of a chore :p) I've ordered a Corsair HX750W. Let's hope it can stop watching videos making me feel like my computer is about to explode :p

Update
My HX750W turned up. The great news is that it makes no buzzing at all under GPU-stress (that I can hear). It also came in a velvet bag :o

But the bad news is some of its cables seem like they could have had a bit more thought put into them. The combined EPS12V/ATX12V 8-pin/4-pin connector splits off right at the end of the lead, so if you split it you end up with a stiff V shaped thing, which is likely to get in the way of something. It doesn't look like the image on Corsair's website, but like this which I borrowed from Newegg. In my case, connecting it split resulted in it blocking a PCI Express slot - though luckily I had enough space to connect it with both bits joint together. If it split a bit further up the lead, it wouldn't have be a problem. Though my old Antec Neo HE just had separate 4-pin and 8-pin leads.

The other problem I had were the SATA power leads. The cable is flat and not rounded, it is quite stiff and ends up bending outwards quite far which is no good if your drives are close together and you don't have a lot of clearance (which I don't). The way they bend out also gets in the way of using two sets of leads in the same run to work around the clearance issue because the first lead gets in the way. I had to rearrange my drives to get around this. My old Neo HE did far better - it had sleeved, rounded SATA power cables which I was able to bend sideways out of the way.

Let's put the View menu before the Edit menu

Well, it has been annoying me at least. I might be alone on that though...

Windows 7 RC: minor follow-up

I've discovered that right clicking on various parts of the taskbar/jumplist brings up some menus I find useful - so that's a great help.

However, quite a few times when I only meant to click on the title bar of a maximised window I've accidentally restored it - maybe some use of SM_CXDRAG/SM_CYDRAG would be an idea here (?)

Also, with regards to the UPEK Protector Suite issues, to be fair to them after some correspondence they've basically fixed most of the problems, so I have a working fingerprint reader now ;)

The only other gripe I have at the moment is that there is far too much stuff in the left pane of file open & save dialogs. I was quite happy with the way it was in Vista where I had shortcuts to everything I needed.. but never mind.

Credit card fraud

Recently, my credit card was used fraudulently online. It seems to be part of a recent spate of fraudulent transactions relating to people who have shopped at a particular online retailer at some point.

The fraudulent transaction itself went through a third party processor, but there was a merchant name alongside it in my credit card statement.

A quick Google search shows only one website of significance going by that name. It is a website selling some kind of web hosting related software. A quick glance at the website reveals nothing too untoward. However, a bit of further digging and some oddities crop up - such as the 'Buy Now' link just takes you back to the homepage. There are a few other inconsistencies as well.

At this point I was suspicious, and thought that perhaps this website was just set up in order to get an account from the payment processor. Presumably, he would use this account himself to process fraudulent transactions. (I do not profess to know how likely this theory could be.)

I looked up the domain name's whois details, which again threw up some items of interest. In particular, a web search of the name of the person the domain was registered to indicated that he has apparently been involved in some eBay fraud too, and threw up his bank details.

It was at this point, I felt like I had too much information (again). What do I do with all this information? Is there actually someone that would care about it? I may never know.......

Some Windows 7 RC impressions...

I gave in and upgraded Windows Vista SP2 on my laptop to Windows 7 RC. (Moreover, I plan to buy a new laptop around when RTM is released anyway, so I hopefully don't need to worry so much about the update from RC to RTM).

Some observations I made:

Installation/Upgrade from Vista

• It asked me to uninstall PerfectDisk (2008) before it would let me continue. Fine I did so. But it still persisted to ask me to uninstall it after it was uninstalled. I had to have a look in the Windows 7 setup log to find out what it was picking up. PerfeckDisk is one of those applications that likes to copy its setup files to a separate folder within the Program Files folder. Apparently Windows 7 was checking every binary in the Program Files folder (or something), and so was picking those up.. Hence, removing those setup files overcame this. (There was some other strange things about some of the apps that were detected, but the other ones didn't block the installation at least).
• The upgrade took a fair while - about 3.5 hours. It spent a lot of time gathering and then restoring system/user files, settings and programs. I won't question it...
  
• After upgrading, Kaspersky decided to forget its "activation code" (i.e. licence key). Whoever the fault lies with, uninstalling it was easier than bothering to find the licence key.

Windows Explorer

• Where are the sort header buttons in Windows Explorer gone that were in Vista (in views other than details)? It doesn't seem that I'm alone in missing them!
• The status bar doesn't display file sizes anymore, as it is apparently superseded by the details pane. Yet the details pane is slow to update, and doesn't display file sizes when you have more than 15 files selected - a bit of a nuisance.
• They seem to have a new list control here, unfortunately it has does have some strange quirks.
  

Biometric Support

Initially it seemed pretty nice, as that when combined with the new UPEK Protector Suite software, fingerprint recognition at the logon screen when the computer was locked was really quick compared to with the old software on Vista/XP. But, it mostly has stopped working after having my laptop on for a few days (generally, I use standby and hibernate rather than shutdown). Most of the problems are probably down to the UPEK software, though. Anyway now that my laptop has not been restarted for several days we have:

• Doesn't recognise fingerprint swipes for most of the sessions after coming out of standby. Maybe related to this message in the event log: "S1: Fingerprint sensor device communication error."
  
• Windows Biometric Support service crashing on the occasion in some UPEK library
• The issue in the below post with the UPEK software leaking things at stupid rates
  
• This WTF message from the UPEK software (followed by it not working):

On reflection, the problem here seems to be general suckage of the UPEK software.

Update: Amusingly, UPEK have apparently found me judging by the comment left on this post :p

"Superbar"

It has its positives and negatives:

• What is really nice is the space it gives. I generally have lots of things running so I do appreciate this.
• But when you have multiple windows of an application open, it can be quite hard to navigate, especially with a touchpad. I may have five explorer windows open - if the Explorer icon is in the bottom left of the screen the thumbnails will be weighted to the right of the icon. Then to copy and paste a file from fifth window to the fourth with the touchpad, it involves a lot of work. In this respect, it would be easier if the thumbnails were stacked vertically, as you could move the mouse in a straight vertical motion. Or perhaps if the Alt-Tab menu had labels underneath each thumbnail.
  
• I did also think one of the reasons of web browser tabs was to reduce taskbar clutter - showing them as separate 'sub-windows' seems to negate that somewhat to me. I often keep 20+ tabs open in Firefox, so it seems this would quickly fall apart. It also breaks the simple act of clicking on the taskbar icon to do a simple minimise/restore. Though, I appreciate that sometimes it is quite useful.
  
• Why can't the jump list appear next to the mouse when you right click instead of above the taskbar? Again more unnecessary mouse movements. The jump list also makes doing something like 'Open file location' a chore (compared to the old quick launch) as the command does not appear when you right click on the application entry in the jump list (you have to go through properties).
• Also, the position of 'Show desktop' seems a failure to me. I often blindly click there when I instead wanted to show the calendar. And when I want to show the desktop, I find myself near the start button - it seems like a better location also because icons on the desktop are usually located on the left.
• I do feel that a medium option for the size of the taskbar icons would have been nice, alongside the existing small and large options.
• Why is the text used in Jump Lists weirdly blurry? Comparison below (top is main pane of start menu, bottom control panel jump list). It does depend on the angle I look at my laptop display, actually. If you zoom in closely, the reason it might look quite different is obvious.

That's enough for now...

Don't you love brilliantly written software?

No further comment. Well, except that that is UPEK Protector Suite 2009 build 5.9.1.5395. Maybe I will try the feedback form....

When in doubt, look closer to home

I saw a string of peculiar messages in the Windows Event Log on my laptop which at first sight looked quite alarming:

Windows (1608) Windows: A request to read from the file "C:\ProgramData\Microsoft\Search\Data\Applications\Windows\Windows.edb" at offset 56401920 (0x00000000035ca000) for 16384 (0x00004000) bytes succeeded, but took an abnormally long time (13930 seconds) to be serviced by the OS. In addition, 0 other I/O requests to this file have also taken an abnormally long time to be serviced since the last message regarding this problem was posted 30928 seconds ago. This problem is likely due to faulty hardware. Please contact your hardware vendor for further assistance diagnosing the problem.

I was skeptical that my hard drive had issues. But (slightly alarmed) I thought I'd give the message the benefit of the doubt and investigate further.

First thing to check was, are there any other signs of hard drive issues? S.M.A.R.T.? No problems. Chkdsk? A-OK. Any different class of messages about hard drive/storage related problems in the Event Log? Nope.

So, as sure as one could be that my hardware was OK I took a closer look at the messages themselves. The first thing that stuck out was the times quoted: 13930 and 30928 seconds. Hmm, that's a long time - in fact respectively 4 and 8.5 hours!

The plot thickened: the messages seemed to be following a pattern - they were all straight after resuming from standby. Yes, it was painfully obvious by now but those 4 / 8.5 hours were how long the laptop was in standby. I didn't need to contact my hardware vendor to work that out, either.

Conclusion? I don't know: probably that if you are going to blame hardware, make damn sure you're right in the first place. (I can only begin to imagine what the conversation with my hardware vendor might have actually gone like...)

Repeatedness

If A does something a handful of times, it is described as repeated, and yet when B does it several times as many as A it's not worth mentioning. Strange, isn't it?

Kaspersky vs. Visual Studio and the rest of my computer

I thought I would try Kaspersky Internet Security 2009, since someone else in this household uses a certain bank gives out free licences.

I did face some issues though:

• I only wanted the anti-virus part of it: not the firewall and other bits. Of course the options in the installer are so obtuse, it became a game of trial and error.
• It refused to install with Spybot Search & Destroy installed - well actually, it said it would be automatically uninstalled if you continued. That is despite the fact I didn't have any of the real-time scanning features of Spybot enabled.
  
• After installing Kaspersky, I noticed that in the Windows Vista Network & Discovery Center, network discovery was set to 'Custom' and couldn't be turned on. A little bit of digging around, I found out this is because Kaspersky takes it upon itself to disable the DNS Client service.
  
• The next problem was that I was getting build errors in Visual Studio. I have VS set to generate .cod listing files for my projects, and it was randomly giving access denied errors on those. Disabling the real-time protection sorted that problem: it seems hard to believe modern anti-virus software has this kind of problem, who would want real-time AV protection if it randomly breaks your applications?
• I then found out that it took it upon itself to break Internet Explorer 64-bit.
• Finally, sometime later I realised that Autoplay/Autorun had stopped working for all types of media (USB flash drives, CDs, ...). I didn't link this to Kaspersky initially, but after some (virtual) digging it seems like it had something to do with it. If you are wondering, the fix was to sort out the NoDriveTypeAutoRunregistry key (which had been set to 0xFF) described in this Microsoft KB article in the "How to selectively disable specific Autorun features" section. I think this was followed by a log-off & on (or otherwise maybe a reboot).
  

Not to worry - even if it was a journey of fun, it didn't cost me anything.

Nokia N85: The dodgy earpiece

I quite liked my N85. Then I read about how various recent Nokia phones made before Feb 2009 had a problem with a faulty earpiece part. Naturally, that followed with my earpiece breaking: it would cut out and in if you tapped the area around it.

So I thought I'd try and get it fixed. My first thought was to take it to one of the Nokia Care points which mainly seem to be Carphone Warehouses and also some random dealers. I tried to call the local Carphone Warehouses on the list a few times. However their repair sections never bothered answered the phone every time I called. Like I'm going to leave my phone with you, then.

So I thought I'd post it off to the postal repair service instead, which seems to be operated by a company called A Novo. My phone came back about a week later which seemed reasonable. Unfortunately as my gut expected in the first place, they screwed my phone over. The earpiece doesn't work at all now - because they fitted it upside down. In fact you can see the lettering on it through the little hole for the earpiece - which after comparing with some photos on the internet seems to indicate it is still the faulty part!! Even worse, my previously beautiful OLED display now has several clusters of brightly coloured pixels. If it's just trapped dust or not, I'm not sure.

Needless to say I wasn't pleased with all of this (!)

Resolution: TBC....

Unsubscribing from e-mail newsletters

I thought I'd try my hand at this today and just found:

• It slightly annoying when you are sent an e-mail confirming that you've unsubscribed (hello, I just told you I don't want your e-mails!)
• Even worse, you're asked to log into an account that you never created in the first place. My best guess here is that I registered to some other site owned by the same company. Though their 'password reminder' e-mail still hasn't come through.....
  

Why do people fail at handling WM_MOUSEWHEEL properly?

When handling this message, one of the values you receive is the value which expresses "the distanced the wheel is rotated" (or rather, was). I will call it delta1 to avoid confusion. Basically, this works on a scale where a delta1 of [fixed value called WHEEL_DELTA] is equal to [system wide setting] units (lines or whatever). With respect to that fixed value WHEEL_DELTA, it says on MSDN:

"The delta was set to 120 to allow Microsoft or other vendors to build finer-resolution wheels in the future, including perhaps a freely-rotating wheel with no notches. The expectation is that such a device would send more messages per rotation, but with a smaller value in each message. To support this possibility, you should either add the incoming delta values until WHEEL_DELTA is reached (so for a delta-rotation you get the same response), or scroll partial lines in response to the more frequent messages. You could also choose your scroll granularity and accumulate deltas until it is reached."

Unfortunately, many people seem to have glossed over that, and even the rest of the content on the page.

Some of the problems I have faced when using a smooth Microsoft mouse wheel:

• Applications scrolling in the same direction whichever way you turn the wheel (unless you turn it exceptionally fast). I would guess the bugged code in question would be something like:

  scroll_direction = delta1 >= WHEEL_DELTA ? UP : DOWN

  Microsoft's Performance Analyzer fits this category..
• Applications not scrolling at all unless you turn the wheel really fast. The problem in this case would be the application ignoring |delta1| values less than WHEEL_DELTA. The new PerfectDisk 10 fits this one.
  
• Applications scrolling erratically/"too much". The problem in this case is that they only check the sign of the delta1 value, and ignore the magnitude.
  
• Applications lacking common sense with mouse wheel scrolling. For example, on its rotating main menu Windows Media Center generally scrolls several entries at a time on when turning my mouse wheel.
• Not so much a bug but a nuisance nonetheless: applications accumulating delta1s until |delta1| >= WHEEL_DELTA is reached, when they could in fact have scrolled in smaller increments earlier. This affects many of the lists in the shell on Windows Vista, though I noticed many of those have been changed in Windows 7 beta.

(Click here if you don't know what my vertical bars mean)

Wow

How much misguided crap can a company spew?

Somone listened!

Probably not to me though.

The 'Internet Explorer' Windows Explorer warning I complained about some posts below now looks like this in Vista SP2:


Not only did they remove the Intenet Explorer references and make it make sense, they also changed the icon I complained about :P

I give up. (More flash/hard drive woes)

I ended up sending back the Samsung SSD, and bought an Mtron MOBI 3500 instead. (After a bit more research I found out that the Samsung was somewhat better, but never mind it was too late by then). Installed Windows Vista on it etc., it did what I wanted.

Today I installed the Windows Feature Pack for Storage 1.0 update and restarted my computer.

I was greeted by the lovely message:

DISK BOOT FAILURE, INSERT SYSTEM DISK AND PRESS ENTER.

WTF sprang to mind. I really wanted to blame the update I just installed, but this is at BIOS level so I will have to put it down to a coincidence.

I restarted, checked everything was set-up/recognised OK in CMOS setup and tried again. This time it booted like normal. Clearly I will have to keep an eye on this...

Buying things is difficult

After my problems with the 1.5TB Seagate 7200.11, I did some investigating and just kept finding other people with problems of increasing amounts of reallocated sectors. Seemed like enough reason to stay away from them permanently.

The WD 1TB black seemed like a good alternative, but in the end I somehow ended up deciding to buy a Samsung 32GB SLC SSD. It was the best price/performance compromise for an SSD, as the cheaper MLC ones seem to just suck. (Don't ask me how I decided that a 32GB drive would do instead of a 1.5TB one!)

The drive came, I installed it and booted up to my existing Windows Vista install and opened up disk management. From all the other hard drives I had had recently I was expecting it to prompt me to initialise the drive. It did not, strange I thought. It did not have any partitions but my suspicions were aroused. How it decides to prompt for this I am not sure - maybe it looks at the boot sector.

I used some S.M.A.R.T. software to check the power on hours and count of the drive and was surprised to see the drive had already had a few hours uptime and a bit over 10 power-ons. Could it be from the manufacturer? No, surely they would reset the data if they did even do any tests that would alter them.

So I proceeded to make an image of the drive and use some data recovery software. I discovered someone had installed Windows XP, some Custom PC benchmark program, Crysis and GIMP. Their Windows user name was SSDTESTS!

The dates on the files where long after the drive was made so clearly someone has decided to use the drive shortly to run some benchmarks. I suspect it may have been someone at the store I ordered from - it doesn't really seem like typical customer behaviour!

My problem is now that I have too much information. Does it really matter someone used it for a few hours? Do my principles outweigh the hassle of returning the drive and getting another one? Arrrgghhh!!!

It's usually quite interesting reading the Microsoft User Experience guidelines

...because they seem to like pointing out things that are actually wrong in Windows! Here is one I saw the other day:

Incorrect:


In this incorrect example, the use of glass is distracting. A plain window background would be a better choice.

(It made me laugh, anyway... :P)

This is what happens when you stick to hibernate instead of shut down...

...and keep Visual Studio and Firefox permanently open:

First entry is biggest WTF. It is a part of Visual Studio. What it is doing with 785MB committed memory (~45MB private working set (!) ) is a mystery (well, the problem is probably that it is doing nothing with it..). That is Visual Studio 2008 SP1 also. Maybe if Microsoft included this column by default in Task Manager some application developers may pay more attention to it.. (!)

I had a lot of tabs open in Firefox, but nevertheless it does seem to feed on GDI objects. System-wide there was about 8.7K GDI objects in use... which is the kind of point where I seem to face rendering glitches. Such as black screens in place of UAC prompts. It gets pretty annoying. (Not entirely sure if this is related to GDI resource usage actually.. possibly not.)

The other process using over 1K GDI resources in sidebar.exe. Must be a leak since it doesn't do anything particularly different after some weeks than shortly after boot-up (where it is using about 94). I will axe the only 3rd party gadget loaded and see if that makes any difference..

Yes, it will always be "explorer.exe".

One of the most annoying issues I have with applications when running under Vista 64-bit, is when applications (or rather their developers) decide that it will be a good idea to call explorer.exe directly for their implementation of "Open file location"/"Open containing folder". Apparently Shell APIs do not exist, or maybe they are scared of GetProcAddress et all if they actually care about Windows 2000 support.

Not only does calling explorer.exe directly work like ass, but generally when 32-bit applications decide to do it, it launches an instance of the 32-bit explorer.exe rather than the 64-bit version (I wasn't referring to this in the linked post though). For me I have this issue with Firefox and uTorrent. As a result, all of your 64-bit shell extensions will be unavailable. For me that includes 7-Zip and my anti-virus - the things I usually want to use on files downloaded with those two applications. Thankfully this was at least fixed in foobar2000. [Update: Seems to also be fixed in Firefox 3.1]

Accumulated installer issues
On a vaguely similar note, I seem to remember (some years ago) some retarded setup applications doing something like extracting a file called setup.exe into the temp folder, but then running an entirely different setup.exe from the (downloads) folder where the original executable was. Or something like that anyway.

The other similar installer issue is when (self-extracting) installers - such as most of nVidia's - decide the drive you wish to extract them to is the C: drive - even when it does not exist or is not the system drive. Who knows, maybe they fixed it by now.. I wouldn't bet on it though.

Installers launching the installed application with admin privileges when the installer is running under UAC elevation also seems to be something people have gotten wrong repeatedly in the past. Let's hope not so much in the future.

Another OS bug
It doesn't warrant a new post but I finally tracked down another issue I was having with Windows (Vista). I had downloaded an executable file, moved it into a sub-folder of Program Files, and created shortcuts everywhere for it. Now, whenever I ran those shortcuts, or even the executable directly it would always warn me about the file coming from another computer. Even when I unchecked the box about always prompting for that file. Eventually it annoyed me enough to investigate. I opened the properties for the file, and tried clicking the 'Unblock' button. But it didn't stop the messages, and the button came back next time you opened that page! At this point I could work out what the problem was (two-fold):
1. Vista needed elevation to unblock the file (as it was in the Program Files folder) but failed to prompt for it.
2. It silently failed when it couldn't unblock the file (and temporarily acted like it succeeded).

There is probably a few ways to work around this - I don't remember which one I used, possibly moving the file out of Program Files, unblocking it, and moving it back. But there sure is a couple of sloppy things on Microsoft's side here.

Nokia N85 WiFi - "No gateway reply" with Thomson routers (also E71, N79, 5800, ...)

If you have one of the recent Nokia phones (N85, E71, N79, 5800, ...) and a recent Thomson / SpeedTouch / BT Home Hub router you may face the "No gateway reply" problem when using the WLAN. By trial and error, I found that this can be resolved by disabling WMM on the router via CLI. So it seems all of the phone's traffic is being QOSd away. Whose side the problem is on I don't know; they are both WiFi certified for WMM of course. An N95 (which doesn't have WMM support) is unaffected either way. Just thought I'd save someone who may come across this via Google the hours of pain I went through..

Update
Someone actually did find this apparently judging by the comments. To access the CLI you normally use Telnet. Open a command prompt and type "Telnet 192.168.1.254" (if you changed the default IP... you probably don't need these instructions.) (Note: On Windows Vista you have to install Telnet client first, under Programs and Features in Control Panel). Enter your username and password. (For unbranded routers the default username is Administrator, case sensitive, with a blank password. Things may be different if your router is ISP branded.) Finally run these commands:

:wireless qos config mode=disabled
:saveall

Verdict on replacement Seagate ST31500341AS

It will also be going back, for a refund this time. This one hasn't completely died like the other one, but it makes a click every time it powers up (the same type as the other one did repeatedly when it died), and the reallocated sector count is increasing slowly (on ten so far.. ten too many for a 2 week old drive). I didn't bother installing an OS on it this time.

I don't know if the packing/shipping of the retailer has anything it do with it. It came in a very tight air pack thing (the type where you would put the object inside it and inflate it) and that itself was loose in a large box.

I also found out there are two versions of this drive, ST31500341AS and ST31500343AS. More often than not it seems the former has SD1x firmware version whilst the latter has SD3x firmware. Not much info around on the latter model.

How many bugs do window scrollbars (APIs) have with visual themes enabled?

A never ending amount, evidently. Amusingly they go away if you disable visual themes for the application.

Retrieving data from the dead hard drive

The process I followed is relatively funny really:
1. Unplug hard drive SATA power lead & boot into Windows. Wait until it has finished loading.
2. Connect SATA power lead, copy as many files possible (using Robocopy) before hard drive starts persistently clicking loudly and stops responding (in practice was about 2.3GB worth of files).
3. Remove SATA power lead, wait a little bit until the drive spins down. Then go back to step 2.

That was fun, I can assure you. Handily Robocopy does not recopy files that is already copied sucessfully when you re-run it with the same command, so it was actually very useful.

I also tried formatting the larger partition after I had recovered the files, which of course forces all handles to the partition to be closed. It did about 70GB before it started clicking, which tells me the background things Vista was doing were not helping when I was copying files off the drive. BUT if you have a look around on the internet this drive appears to have issues when running under most other operating systems for some reason (the firmware issues, apparently). Booting a command prompt through the Vista DVD would probably normally be a good option - but I wouldn't have been able to address the >1.1TB part of the drive without loading the new SATA controller driver. It could most likely be done but I just wanted to get my data back ASAP, and I didn't think of this at the time. If my replacement drive goes the same fate I may be trying this method, however.

The most ridiculous thing about this was that I managed to check the S.M.A.R.T. values for the drive whilst I was copying my data from it. And NONE of the values were below their thresholds! It had been through so much clicking at the point, I was expecting something bad, but nope.

Computer woes continued

It seems my suspicions of the nVidia SATA controller driver being bugged were in fact correct. So if you are using the Seagate ST31500341AS or another affected drive (one with enough sectors) on an nVidia chipset, make sure you have the latest SATA controller driver. (It may have been designed like that rather than being a bug, but then it is poor design instead anyway).

As it turns out, it seems my hard drive has let me down afterall. Since today, shortly after logging in to Windows the drive stops responding (well, any app attempting to access the drive does) and emits a continous ticking sound. Nasty. The drive is only two weeks old :/ And, it passes the drive short self test in SeaTools (!) All seems very odd. There are a few similar reports in the reviews on NewEgg, and I've read a few reports that said Seagate themselves have said the SD17 firmware has issues. But the drive was working great up until now so I don't know, I've reluctantly gone for a replacement, so we will see what happens. Some more googling reveals there is in fact a new SD37 firmware. For now, I need to try and get my data off the thing :/

General computer woes

This is what happens when I decide to upgrade some parts of my computer:

1. Seagate 7200.11 1.5TB hard drive

It turns out the nForce 590 (AMD) SATA driver on Windows Update (version 5.10.2600.998; it was a new install of Windows Vista, the previous version included with Vista SP1 also didn't work) has some issues with this drive and/or 1.5TB drives in general. Windows Experience Index benchmarking fails ("Cannot complete assessment"); running the command-line WinSAT tool in verbose mode reveals some scary errors in the hard drive benchmark (when it reaches close to the end of the disk).

I was doubtful something was actually wrong with the drive, and things seemed OK with SeaTools etc. I ran a long drive test in SeaTools DOS overnight just in case. To my annoyance, when I returned in the morning SeaTools had decided to quit to the screen where it tells you how to read the log. Checking the log only showed a start time of the test, and nothing else, so no idea what happened there.

Moving on, the next thing I decided to try was booting the previous install of Vista and running the WinSAT tool against the 1.5TB drive. Strange, no errors. So I wondered what could be different and the only thing that sprung to mind was possibly the SATA controller driver. So I checked and indeed, the new install was running a fairly older 5.10.2600.998 version compared to the 10.3.0.42 version on the old install which I had gotten from the latest nForce driver pack.

So I proceeded to install the latest nForce pack on my new install.. and for some reason it didn't want to install the SATA drivers. So I did them manually through device manager. Finally, that cleared up the WEI/WinSAT problem. No idea if there was any other problems evident as Windows was installed to a 300GB partition.

2. Update motherboard BIOS (M2N32-SLI Deluxe) to prepare for new CPU.

My current BIOS didn't support the CPU I had ordered so I had to update the BIOS in preparation. Given the past woes experienced in updating the BIOS on this board, I had put this off until now.

Rightfully so, it seems. Updated BIOS to version 2101... no boot, just graphics card fan whirring at full speed (?). Reset CMOS and it boots again. A bit of investigation and hassle and it appears enabling SLI memory support causes this (which was fine in the previous version). It may be related to the 2.2V voltage set in the EPP profile of my RAM, but anyway it was working fine before. Several people have reported the same on the Asus forums. Anyway I manually set the timings at 2.0V and that was stable.

3. Actually install new CPU.

Well, physically installing the CPU was easy. But it was running hotter than I expected for a 65W CPU. A bit of investigation and it seemed that the CPU core voltage (as reported by the motherboard) was a bit high at 1.39V. The CPU was rated at 1.30/1.35V (side question: what is the slash meant to mean here? The CPU is AMD ADO5600DOBOX). So I tried manually setting it at 1.35V, but it was still reported at 1.39V. So I then tried seting it at 1.30V and it was now reported at 1.34V. It shaved about 5 degrees C off the reported temperature so a result it seems.

Conclusion: I knew it already, but this motherboard sucks.