Expanded view is much better. Tell me about it!

But why doesn’t the damn website remember you switching to expanded view yesterday? Because it just doesn’t offer the functionality.

I am about to change that and rock your world.

The solution: User CSS

Just create a file called youtube.com.css, or something like that, and put this code in it:

div#watch-player.flash-player,
div#watch-video {
	width: /*854px*/ /*970px*/ 1280px !important;
	height: /*510px*/ /*580px*/ 765px !important;
}
div#watch-sidebar {
	margin-top: 10px !important;
}
div#watch-video-container {
	background-color: #555 !important;
}

To elaborate on those values:

854 x 510 pixels is the default size of YouTube’s Expanded View
970 x 580 pixels is the full width of the YouTube website
1280 x 765 pixels is true 720p size and will load videos at 720p by default, because of the player’s size, ZOMG!

You can (un)comment any set of values you like, or choose your own. Be sure not to make it bigger than your screen is capable of showing, because that would be a waste.

As for putting this file to use on the YouTube website, I refer to the internet.

Some visual examples

(because people’s attention span isn’t getting any better any time soon, they may even have stopped reading after the first sentence – not because I’m a bad writer, oh no):

The video player at default size

The video player set to Expanded View

The video player at full website width

The video player at full 720p width

No, what you actually have to do is apply your personal changes to the default styles in any Writer document, then delete all contents and go to File - Templates - Save Template. This file will be saved as a template for future use as a default template, but you need to make it default first.

Go to File - Templates - Organize. Double Click My Templates and select the template you saved just seconds ago. Click the Command button and click Set Default Template.

Now close Writer and reopen it. If you look at the styles by pressing F11 and modifying them, you’ll see they are exactly as you changed them in your previous document, and they will stay that way for eternity!

The actual source of this solution is found here, but I just saved you the trouble of looking even further.

Sometimes it’s a pain to add 32-bit drivers to the Additional Drivers section when you install a printer on a 64-bit Windows machine and share it in your network, because of compatibility with the 64-bit drivers which needs both bit-versions of the drivers to go with the same name.

When you, for example, install 64-bit drivers through Windows Update, the drivers offered on the manufacturer’s website will not necessarily call your printer by exactly the same name. That’s a no-no for Windows, and your favorite OS will stubbornly refuse to accept those drivers.

Now you may think ‘Well, why can’t I let Windows use the 32-bit drivers that are installed using Windows Update?’ Very clever, but you cannot let Windows Update install 32-bit drivers on a 64-bit machine.

But going down that same path on a detour, it is actually possible to achieve the result you thought you’d achieve, but by having to go through some extra steps and having a 32-bit Windows machine at your disposal.

Solution

For this solution to work, you will need to have both 32-bit and 64-bit Windows installed on one or several computers and be able to connect the printer directly to both versions of Windows.

First step: Install printer on 64-bit Host

This is easy: connect the printer, turn it on and install the drivers through Windows Update. Let it finish, then share the printer in your network. Do not yet attempt to add Additional 32-bit Drivers.

Second step: Install printer on the 32-bit Client locally

Roughly the same as the previous step: connect the printer to the 32-bit Windows machine, turn on the printer and let Windows Update check for updates and let it install the printer driver.

Third step: Locate the 32-bit printer driver on your 32-bit machine’s hard disk

This is more advanced than previous steps. Luckily there are constants we can hold on to to guide us through this step. The directory you must open in Explorer is C:\Windows\System32\DriverStore\FileRepository\.

You will see a lot of directories with cryptic names glaring back at you. One or more directories will be of importance in this step, and those can be identified by sorting the directories by Date Modified. Simply look for the directories that are modified at the time you let Windows Update run to install the printer drivers. Select the directories that fit the time-tag and Copy or Share them.

(If you cannot remember the time you installed your printer, you can always search for the correct directories knowing that a directory of a printer-driver starts with prn, followed by two characters indicating the printer’s manufacturer. For example: in case of an HP printer, it will be prnhpXXX.inf_x86_neutral_XXXXXXXXXXXXXXXX, but in case of a Canon printer it will be prncaXXX.inf_x86_neutral_XXXXXXXXXXXXXXXX.)

Do the same for this directory: C:\Windows\System32\DriverStore\FileRepository\ntprint.inf_x86_neutral_XXXXXXXXXXXXXXXX. (all X’s are random numbers or letters, there will be only one directory that starts with ntprint.inf_x86)

Fourth Step: install the Additional 32-bit Driver on your host

Now back on the 64-bit host on which the printer will be shared. Go to the printer’s properties, tab Share and click the button Additional Drivers. Now check 32-bit Drivers and press OK. It will prompt you a dialog to browse for the .INF file that contains the 32-bit driver of the printer you shared.

The correct .INF file will be inside one of the directories you just copied or shared. Just try them all out until one works.

Then Windows will ask for correct the ntprint.inf file. That one can be found in the ntprint.inf_x86… directory you also copied or shared. The file will be accepted and the installation of the 32-bit driver will continue and finish.

Fifth Step: Enjoy how things work!

I know I did. I installed all shared printers (both a Canon inkjet and an HP laserjet) on all computers I could find in my home, all with great success.

More details on specific cases

The installation of the HP Laserjet on the 32-bit Windows machine created just one directory in the FileRepository and was called prnhp001.inf_x86_neutral_XXXXXXXXXXXXXXXX. That was easy.

More difficult was the installation of the Canon printer, which created a whole series of directories, all of them starting with prnca00, and only one of them (to be precise: prnca00n…) containing the driver of the printer I tried installing. Worse still was the Core Driver‘s .INF file which was to be found in yet another directory and ultimately turned out split over two directories: prnca00x and prnca00y. A lot of tedious searching, but in the end everything works as it should, and that’s what’s important.

After some research, I found the cause to be the sound chip power management on OS-level. It is all explained here.

To recapitulate:

Sorry I should have posted more detail.
The site that I got my information from was: Audio Device Class Inactivity Timer Implementation (Windows Driver Kit)

My “IdlePowerState” was set to 04, which is D3Cold in their terms; I changed it to 03 (which I assume is D3Warm, and which I also assume is the pre-Windows 7 setting); it made no difference. Since this is a desktop and not a laptop, I ignored the ConservationIdleTime setting (since that applies to running on battery power), and I increased toe PerformanceIdleTime from 012c (or in the registry, “2c 01 00 00″) which is 300 seconds, to “2c 0f 00 00″ which is about 67 minutes. and so far that seems to have fixed it. I assume if I’m idle for over 67 minutes and then I get an email, it’ll “pop” again, so my choices are to change the IdlePowerState to 00, or increase the PerformanceIdleTime to an even bigger value. You may find many copies of these values in the registry, remember to only change the one under hkey_local_machine system CurrentControlSet.

Good luck. By the way, funny coincidence, I just bought my wife a dv7t, it doesn’t have this problem.

For those reading this anew, the sound is as if someone were plugging the speakers into the jack.

So, basically it says to alter the value in the registry key to set the time-out value to be a lot higher.

I have put the value “2c ff 00 00″ in this key: \HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Class\xxxx\yyyy\PowerSettings\PerformanceIdleTime, which means that in Performance Profile the stand-by mode will only be initiated after a time of 0xff2c seconds of making no sound whatsoever, which means about 18 hours in human language.

In the key \HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Class\xxxx\yyyy\PowerSettings\ConservationIdleTime, I’ve put the value “1e 0f 00 00″, which means 0xf1e seconds (~= about an hour) of required idling before the sound chips kicks into stand-by mode in Power Saving Profile.

(Note: the xxxx and yyyy are GUIDS, large identifier strings chosen by Windows. To find the right one, you need to look up your audio chip in the Device Manager and look at the Properties, tab Details and there Device class guid)

If you looked closely at the default values in your registry, you might have noticed an idle time of 300 seconds, as stated in the quote above. This means that if the audio chip didn’t play a sound for 5 minutes, it’s kicked into stand-by mode, which means that whenever a sound is played from that point onwards, it will start with a *pop*, and that’s annoying.

Facebook Social Plug-in integration as seen on www.engadget.com

The problem in Opera is that when you visit a website that implements the Facebook Social Plug-in, Opera goes berserk and goes off and on loading for an eternity after the website itself finished loading, which is shown in the lovely screenshot below. The result is high and constant processor usage and a bogged down system.

When this is examined using Opera Dragonfly, it becomes clear that Facebook keeps sending AJAX requests that get rejected by Opera. The reason for this apparently is some incompatibility with the cross-domain counter-measures of Opera, or broken browser-sniffing inside the plug-in’s javascript code.

Opera Dragonfly show repeated attempts to set a cookie originating from a facebook domain.

A detail of such an attempt showing the origin being a facebook domain.

At the moment there are two ways to stop the constant attempts which appear to you as if the site keeps loading: a fix or a workaround.

Workaround

A possible workaround may be to block the plug-in’s domain and keep it from being loaded by, which is explained on the Opera forum.

The workaround consists of adding these two URL’s to your blocked content:

• http://www.facebook.com/plugins/activity.php*
• http://www.facebook.com/widgets/activity.php*

The biggest problem with this approach is that you cannot use the Facebook integration on the websites that use it.

Fix

The fix comes in the form of an add-on named “AntiBrowsniffer“, and requires Opera 11.

Comments show that the problem is solved:

aaaaaaaaaaaaaaaaaaaaaaa. you fixed facebook connect problem with this. thank you a million i can connect to facebook with external sites now, facebook comment works, facebook games work. incredibile.

Simplest step

Simply try and enable Multicore Rendering in the advanced video options of your favourite source game. If the game is still fluent and without significant mouse lag, then you are set. But if it isn’t, they you aren’t. More drastic measures are in order.

Some ways to try and make it work without much hassle

I noticed that whenever I enabled Multicore Rendering in Team Fortress 2, it would cause insufferable mouse lag. You know, the kind where your cross hair seems to be connected to your mouse with an elastic. You swoosh your mouse to the left, and some tenths of seconds later the cross hair follows your movement.

There is a way to fix this, although it might not be pretty because there is a visual trade-off, and it is by disabling Vertical Sync (or ‘Vsync’). This causes tearing in the image, but it also causes the computer not to wait for a completely filled buffer before outputting that buffer to the screen, therefore eliminating another step in the process that causes lag.

Drastic ways to make it work if it still wouldn’t.

There are a number of things you can try before giving up:

1. enter “mat_queue_mode 2″ in the console
2. update your graphics driver to the most recent stable version
3. restore the settings of your Graphics Card to default values as much as possible

If that still did not make it work, well then, I don’t really know what to propose next.

You know when you play Team Fortress 2 on a 32-player server, and when all of those players are in the same place at the same time, generating a lot of flying projectiles and explosions, the framerate would free fall below the comfort-zone of 30 frames per second – or whatever your personal preference is in an online first person shooter.

Whenever you go look on the internet for a possible solution to this problem, you find that it helps if you enable Multicore Rendering, which is a setting in the Advanced screen of the Video Options. There is a big chance though that when you enable this setting, your ingame experience might worsen, which is the exact opposite of what you were trying to achieve by enabling this option in the first place. “Great.” you might think cynically, and turn it back off again and cope with the slowdowns in crowded areas for the rest of your otherwise fun day, which seems is all you can do.

Or is it?

Great news! I personally managed to get it working in an almost perfect way, at least to me. But first let’s start at the beginning of my encounters with this very weirdly behaving setting.

When I first enabled it, I suffered from enormous mouse lag, the kind where I would think the weapon I control with the mouse isn’t attached to the mouse, but to an elastic which is in turn attached to the mouse. Not very handy when you are trying to blast your enemies back to their spawn with lightning fast reactions and surgeon-like precision. No sir, it isn’t. “To hell with this!” I thought and disabled the setting immediately. A while later I decided I’d have another go at it, but to no avail. The same weird problem I had still bugged me.

But a third time, months after the second one, I again got really annoyed by the way the framerate takes a dive whenever I have a lot of other players in my sight. “No more of this rubbish!” and I googled for everything I could think of that might be related to my problems. So first I found that a lot of other people were also suffering from the bad way this setting was functioning on their computers. But then I was so smart to think that adding the keywords “mouse lag” to the seach query might be a better way to find a solution to my problem.

The Great Discovery

I stumbled upon this thread on the Steam forum. Some people there mentioned the mouse lag would disappear when Vertical Sync is disabled, which was enabled the entire time on my computer, because I fancy the way it makes the experience a lot smoother. The other thing they mentioned was this setting I’d have to change by entering something in the console: “mat_queue_mode 2″, which enables “Queued multithreading” supposedly.

So I changed both settings to the state where they should make Multicore Rendering work like it should. The result was that it was quite nice: the framerate didn’t go down the way it did, but somehow I noticed the game to be more choppy, although the reported framerate was telling me it really shouldn’t. To elaborate on this, at that time I was using an ATI Radeon HD4770 with ATI Catalyst 10.10 drivers on my computer with Windows 7 64-bit installed. In the Catalyst Control Centre, I had set Catalyst A.I. to an advanced level. I don’t know for sure if this had anything to do with it, but to me the choppyness of the way it felt looked like some form of microstuttering which I had read about on the internet.

The last few tweaks

But very recently, Microsoft pushed Catalyst 10.11 via Windows Update, and at the same time I was rumbling inside the CCC 3D-settings panel, and then I thought “Why don’t I disable this override thingy and set the Catalyst A.I. to basic, because maybe these settings were interfering with the Multicore Rendering setting of the Source engine.” So I did all that and then fired up the game, and would you look at that! It was barely noticeable  that I somehow had Multicore Rendering enabled. The game felt very fluent once again. The only thing that might still bother me is the VSync which still needs to be disabled. I tried to enable it again, and although the mouse lag had faded somewhat, it was still a tiny bit there. I can’t live with that. Not in this game. Although the screen tearing is less noticeable now that the game feels a lot more fluent again, it’s still there and I’m sad about that. But hey, now I don’t have the problem of my framerate collapsing like a weak pudding.

Conclusion & tl;dr

So there, that’s my semi-bloggy story about my quest to have this shite of a setting properly working.

For people who couldn’t be bothered to read my great story, I will recap the necessary steps here:

1. enter “mat_queue_mode 2″ in the console
2. disable Vertical Sync in the Video Options – Advanced
3. restore the settings of your Graphics Card to default as much as possible

Now, Up and Atom!

There is a problem, however, and that is the hassle to get the looks of those fonts right. After you figured it all out, which can take a while if you are on your own, it is actually quite simple. But like every “fix” for Ubuntu, it takes a lot of time learning what is causing your problems and finding a solution for it.

How it’s done

There are two steps involved to make it happen:

1. set the system to use Microsoft fonts as the default system fonts
2. make them look good in every application you use, from native Gnome applications to native KDE applications to QT applications and GTK applications.

The first step is done in a jiffy and a how-to is found quickly on the internet. The second step, however, while swiftly performed, was a pain in the ass to figure out completely, because of the variety of libraries a GUI can be built for.

First Step

This is done in the menu System which is located on the upper panel, then Preferences and finally Appearance.

Set every font and its size to your liking.

Second Step

This step is about setting the rendering options for fonts to the correct values. While these options can be set in the window used in step 1, some applications refuse to use those set options and decide to use some default preset values which may differ from the values set in the options window. The clue is to make those applications use the font rendering settings that you chose. “Where do those applications get their settings from, if not from the gnome font options window?”, I hear you ask yourself. Well, there are multiple alternative sources to read font rendering settings from, which are all file based. The files in question are ~/.fonts.conf, ~/.Xresources and /etc/fonts/local.conf and are each the default settings sources for different sets of applications. Mozilla Firefox and OpenOffice.org for example both first rely on the file ~/.Xresources, while qt3 and qt4 applications rely on the ~/.fonts.conf and /etc/fonts/local.conf files.

Let’s start off with the most intuitive source of font rendering settings: the gnome appearance preferences window.

Appearance Preferences

In the same window as in the First Step, you have to set the font rendering options to the correct values, which is done selecting Best Contrast in the lower area of the window. More specific changes can be made by clicking the Details button. This opens the window shown below.

Since it doesn’t really matter you can set font Smoothing to Subpixel or Greyscale, whichever looks best for you. If you selected Best Contrast in the previous window, then Smoothing will be set to Greyscale.

The difference between the two settings is that Greyscale smoothing applies anti-aliasing to the fonts by adding grey coloured pixels to the edges of the fonts. Subpixel smoothing, on the other hand, adds coloured pixels to the edges of the fonts. Which colours depends on the Subpixel Order, which can be set in the same window. Depending on the screen you use, this will give an optimal or a sub-optimal result. Most screens have their subpixels arranged in the order RGB, which means Red Green Blue. Thus when Subpixel Order is set to RGB, then Subpixel smoothing will result in blueish coloured pixels around the left edge and reddish coloured pixels around the right edge of the fonts. While those colours might seem odd when used to smoothen the edges of black coloured fonts, the colours don’t become visible since a subpixel is too small to distinguish with the naked eye from a reasonable distance.

The last setting named Hinting, however, proves to be much more important, since it can make or break the looks of the classic Microsoft fonts like Arial, Times New Roman, Verdana, etc.

~/.fonts.conf & /etc/fonts/local.conf

These files offer an XML based structure for your font rendering settings. Since it is our intention to copy the settings from the Gnome Appearance Preferences window as much as possible, I will just give you the contents of the file I put together:

<?xml version='1.0'?>
<!DOCTYPE fontconfig SYSTEM 'fonts.dtd'>
<fontconfig>
 <match target="font" >
  <edit mode="assign" name="autohint" >
   <bool>false</bool>
  </edit>
  <edit mode="assign" name="rgba" >
   <const>rgb</const>
  </edit>
  <edit mode="assign" name="hinting" >
   <bool>true</bool>
  </edit>
  <edit mode="assign" name="hintstyle" >
   <const>hintfull</const>
  </edit>
  <edit mode="assign" name="antialias" >
   <bool>true</bool>
  </edit>
  <edit mode="assign" name="lcdfilter" >
   <const>lcddefault</const>
  </edit>
 </match>
 <match target="font" >
  <test compare="eq" name="family" qual="any" >
   <string>Calibri</string>
   <string>Cambria</string>
   <string>Candara</string>
   <string>Corbel</string>
   <string>Constantia</string>
   <string>Consolas</string>
  </test>
  <edit mode="assign" name="embeddedbitmap" >
   <bool>false</bool>
  </edit>
 </match>
 <match target="font" >
  <test compare="eq" name="family" qual="any" >
   <string>FreeSans</string>
   <string>FreeSerif</string>
  </test>
  <edit mode="assign" name="hintstyle" >
   <const>hintslight</const>
  </edit>
 </match>
</fontconfig>

The first “match” element is used for default settings, while the next two “match” elements are used to further specify or alter settings depending on the outcome of the test, of which the conditions are declared by the “test” element.

As you can see, the second “match” element tests if the font currently treated is one of the new Microsoft Office 2007 fonts. These fonts have broken bitmap rendering which results in non-smoothed fonts, so it is necessary to use an alternate rendering technique in order to avoid this. Too see for yourself: install the needed MS Office 2007 fonts and use them in an OpenOffice.org document with the second “match” element excluded from your ~/.fonts.conf file.

~/.Xresources

The same is true for the ~/.Xresources file. Just a reiteration of the font rendering settings mentioned previously.

Xft.antialias:  true
Xft.hinting:    true
Xft.hintstyle:  hintfull
Xft.lcdfilter:  lcddefault
Xft.rgba:       rgb

History and background of my personal conquest

I have been doing a lot of research on how Ubuntu fonts are actually used by programs. It turns out there are four major sources of settings and each one is used by a different group of applications. For example: Mozilla Firefox and OpenOffice.org use the ~/.Xresources file, while other applications (most GTK and QTx applications) look at the ~/.fonts.conf file or the /etc/fonts/local.conf file if the gnome appearance settings are ignored.

Another thing I noticed is that while some fonts (like the Free font series) look horrible when font-hinting is set to medium or full, Microsoft fonts (e.g. from the msttcorefonts package) tend to look aweful when font-hinting is set to slight. I used FreeSans as my main system font since Ubuntu 8.04, but I always loathed the looks of the Microsoft fonts at slight hinting. Unaware of the cause of the uglyness of the Microsoft fonts, I blamed Ubuntu and its maintainers for letting a regression of this magnitude happen.

Lukily I have it all figured out and I’m happy to share this knowledge with everyone who is interested.

Further investigation

In the (hopefully near) future I will cover the behavior of certain widely used applications and fonts in relation to all possible font rendering settings and sources, which will uncover the great mysteries of font rendering in Ubuntu and GNU/Linux in general. But for now this is all you get from me.