• |
  • |
  • |
  • |


Be aware that this is not a conventional blog. It will not be updated on a daily basis. It's more of a toolbox with things I encounter during my work or problem resolutions found after some daunting trial and error process. Fruitful reading!


To nie jest typowy blog. Nie będzie aktualizowany codziennie. Jest to raczej przybornik z informacjami, linkami i rozwiązaniami problemów, które napotykam w czasie pracy, często na zasadzie prób i błę­dów. Owocnego czytania!


Ce n'est pas un blog typique. Il ne sera pas mis à jour quotidiennement. C'est plutôt une boite à outils avec des choses que je rencontre pendant mon travail ou les solutions des problèmes qui sont à trouver seulement après une recherche exhaustive. Fructueuse lecture!


Beachten Sie, dass das kein Übliches Blog ist. Ich werde es nicht täglich aktualisieren. Es ist eher ein "Werkzeugkasten" mit den Sachen, die ich während meiner Arbeit antreffe und Problemlösungen, die manchmal nur durch Ausprobieren gefunden werden können. Fruchtbares Lesen!


Esto no es un blog típico. No voy a actualizarlo diariamente. Es más bien una "caja de herramientas" con las cosas que encuentro durante mi trabajo o resoluciones de los problemas que descubro después de una búsqueda exhaustiva. ¡Lectura provechosa!

Jolicloud terminal access (keyboard shortcuts)

Disappointed with the usability of the new Unity interface in the Ubuntu Netbook Remix edition (my screen is not wide enough) I’ve recently decided to switch my Linux netbook OS. My distro of choice was MeeGo, but it had some major issues with my Wi-Fi setup, so unfortunately I had to let it go. I went for Jolicloud (project by Netvibes’ founder).


My first impressions were rather unpleasant ones. Though the general feel of the launcher is good enough, I couldn’t find an easy way to access the terminal (apart from a dozen clicks through local/native apps) . And what kind of Linux distribution is this if I can’t easily access the command line?


Fortunately, the terminal is conveniently accessible through a not-so-well-documented keyboard shortcut that I’ve stumbled upon reading the tips & tricks section in the Jolicloud blog. Just press Alt+F1 to bring up the console. By the way – the Menu key gives access to the standard Ubuntu menu (Jolicloud is based on Ubuntu), which can come in handy too.

Full listing of the shortcuts (from Jolicloud blog):

  • Windows key – main launcher
  • Alt+Tab – application switching
  • Alt+F1 – console access (via classic terminal app)
  • Alt+F2 – run command (application text launcher)
  • Alt+F4 – close application
  • Menu key – Ubuntu style menu

Using those keyboard shortcuts Jolicloud’s usability is acceptable for daily use (from my perspective). It could use some improvements here and there, but it’s much better than UNR 10.10. However, the joint venture of Nokia (Maemo) and Intel (Moblin) in the form of MeeGo looks really promising. So when major bugs are ironed out and more hardware gets supported out of the box it can become one of the most promising netbook (mobile) operating systems around. Just wait till Smeegol (or SUSE MeeGo) comes preinstalled on new hardware. Jolicloud will definitely need to keep up pace.

Why using server.error-handler-404 is bad?

OK, maybe the title is a bit too provocative. More precisely said, “why using server.error-handler-404 directive for clean URL rewrites in WordPress is a bad solution” would be more appropriate.


Switching from Apache to Lighttpd some time ago I used the simplest solution to port my WordPress installation. The quick and dirty way to do it as proposed on numerous occasions (for example chrisjohnston.org or ubuntu-tutorials.com) is OK as an easy fix for most sites – in you virtual host put:

server.error-handler-404 = "/index.php"

This quick trick redirects all traffic to the WordPress core with a single line edit. Brilliant in its simplicity and works fine as a temporary fix. But there’s some nasty side-effect I wasn’t aware of. My wp-recaptcha plugin ceased to show error messages without a reason. Or that’s at least what I thought.


I pinned down the problem to the query string being not set when using this kind of Lighttpd setup (see the environment details for more info) – all additional GET variables are lost this way. The information from Daniel’s blog were really helpful here. Now I’m happily using mod_magnet for this without any problems so far. A bit more to setup, but way more flexible. See the tutorial at longspine.com for detailed information.

DualView on VirtualBox Windows guest


Multiple displays are still scarcely supported in virtualization solutions. Theoretically, VirtualBox supports multiple displays, though it applies only to Windows guests when using VBoxHeadless (RDP access). See the User Manual for the details.

In practice it’s not that trivial to set up, and I had to spend a few days overcoming some problems on the way. My setup is Ubuntu 9.10 (Karmic Koala) host with a Windows XP guest system. One thing which should be done in the first place is to increase the video memory available to the guest.

Problem #1:

On my Linux host I currently use TwinView (NVIDIA’s Xinerama implementation). This means only one big X screen is used that spans all the physical displays. For normal operation it’s quite fine, it’s possible to have windows “in-between” and the internals take care of maximizing to one screen only etc. By the way – the Compiz plugin Put is very handy – just bind “Put To Next Output” to a hot-key to move windows between displays in an efficient manner.

At times, however, this logical separation of the screen fails to function properly. And that’s the case with vanilla rdesktop. The unpatched version will span both monitors when going into fullscreen mode. This way it’s completely unusable.

Solution #1:

Fortunately, this has been a long known bug of rdesktop. Thanks to Rasmus Jacobsen’s post, I was able to redownload the original patch with some modifications and apply it to rdesktop 1.6.0. For redundancy, here are his instructions to applying the patch:

sudo apt-get install devscripts build-essential fakeroot
sudo apt-get build-dep rdesktop
apt-get source rdesktop
cd rdesktop-1.6.0
patch  -p0 < /tmp/rdesktop-xinerama.patch
debuild -us -uc
cd ..
sudo dpkg -i rdesktop*.deb

To get it working with rdekstop 1.6.0 I had to change the line:

+ XGetWindowAttributes(g_display, parent, &win_attrib);


+ XGetWindowAttributes(g_display, g_wnd, &win_attrib);

The full modified version of the patch is here:

--- configure.ac.orig	2008-06-25 23:22:27.000000000 +0200
+++ configure.ac	2008-06-25 23:17:51.000000000 +0200
@@ -18,6 +18,8 @@
 AC_SEARCH_LIBS(socket, socket)
 AC_SEARCH_LIBS(inet_aton, resolv)

+AC_CHECK_LIB(Xinerama, XineramaQueryScreens, AC_DEFINE(HAVE_XINERAMA) LIBS="$LIBS -lXinerama", [], [])
--- xwin.c.orig	2008-06-25 23:50:02.000000000 +0200
+++ xwin.c	2008-06-25 23:49:25.000000000 +0200
@@ -21,6 +21,9 @@

+ #include
@@ -570,6 +573,71 @@
 #define LOUT24(o, x) { *(o++) = x; *(o++) = x >> 8; *(o++) = x >> 16; }
 #define LOUT32(o, x) { *(o++) = x; *(o++) = x >> 8; *(o++) = x >> 16; *(o++) = x >> 24; }

+int g_last_wnd_pos_x;
+int g_last_wnd_pos_y;
+ScreenSize(XRectangle *screen)
+	int screens;
+	int event_base;
+	int error_base;
+	if (XineramaQueryExtension(g_display, &amp;event_base, &amp;error_base))
+	{
+		XineramaScreenInfo *screeninfo;
+		int i;
+		/* Get the Xinerama screen infomation. */
+		screeninfo = XineramaQueryScreens(g_display, &amp;screens);
+		/* Search for the appropriate screen. */
+		i = 0;
+		while (!(screeninfo[i].x_org <= g_last_wnd_pos_x
+			&amp;&amp; screeninfo[i].y_org <= g_last_wnd_pos_y
+			&amp;&amp; screeninfo[i].x_org + screeninfo[i].width >= g_last_wnd_pos_x
+			&amp;&amp; screeninfo[i].y_org + screeninfo[i].height >= g_last_wnd_pos_y ))
+		{
+			i++;
+		}
+		if (i >= screens) i = 0;
+		/* Position according to the present screen. */
+		screen->x = screeninfo[i].x_org;
+		screen->y = screeninfo[i].y_org;
+		screen->width = screeninfo[i].width;
+		screen->height = screeninfo[i].height;
+		/* Free allocated memory. */
+		XFree(screeninfo);
+	}
+	else
+	{
+		/* Xinerama is not in use, default to the XLib screensize call. */
+		screen->x = 0;
+		screen->y = 0;
+		screen->width = WidthOfScreen(g_screen);
+		screen->height = HeightOfScreen(g_screen);
+	}
+ScreenSize(XRectangle *screen)
+	screen->x = 0;
+	screen->y = 0;
+	screen->width = WidthOfScreen(g_screen);
+	screen->height = HeightOfScreen(g_screen);
 static uint32
 translate_colour(uint32 colour)
@@ -1615,17 +1683,26 @@
 	if (g_fullscreen)
-		g_width = WidthOfScreen(g_screen);
-		g_height = HeightOfScreen(g_screen);
+		XRectangle screen;
+		ScreenSize(&amp;screen);
+		g_width = screen.width;
+		g_height = screen.height;
 		g_using_full_workarea = True;
 	else if (g_width < 0)
+		XRectangle screen;
+		ScreenSize(&amp;screen);
+		g_width = screen.width;
+		g_height = screen.height;
 		/* Percent of screen */
 		if (-g_width >= 100)
 			g_using_full_workarea = True;
-		g_height = HeightOfScreen(g_screen) * (-g_width) / 100;
-		g_width = WidthOfScreen(g_screen) * (-g_width) / 100;
+		g_width = screen.width * (-g_width) / 100;
+		g_height = screen.height * (-g_width) / 100;
 	else if (g_width == 0)
@@ -1734,14 +1811,19 @@
 	long input_mask, ic_input_mask;
 	XEvent xevent;

-	wndwidth = g_fullscreen ? WidthOfScreen(g_screen) : g_width;
-	wndheight = g_fullscreen ? HeightOfScreen(g_screen) : g_height;
+	XRectangle screen;
+	ScreenSize(&amp;screen);
+	wndwidth = g_fullscreen ? screen.width : g_width;
+	wndheight = g_fullscreen ? screen.height : g_height;
+	g_xpos = g_fullscreen ? screen.x : g_xpos;
+	g_ypos = g_fullscreen ? screen.y : g_ypos;

 	/* Handle -x-y portion of geometry string */
 	if (g_xpos < 0 || (g_xpos == 0 &amp;&amp; (g_pos &amp; 2)))
-		g_xpos = WidthOfScreen(g_screen) + g_xpos - g_width;
+		g_xpos = screen.width + g_xpos - g_width;
 	if (g_ypos < 0 || (g_ypos == 0 &amp;&amp; (g_pos &amp; 4)))
-		g_ypos = HeightOfScreen(g_screen) + g_ypos - g_height;
+		g_ypos = screen.height + g_ypos - g_height;


@@ -1882,6 +1964,11 @@
+	Window root, parent, *children;
+	unsigned int nchildren;
+	XWindowAttributes win_attrib;
 	Pixmap contents = 0;

 	if (g_seamless_active)
@@ -1895,6 +1982,17 @@
 		XCopyArea(g_display, g_wnd, contents, g_gc, 0, 0, g_width, g_height, 0, 0);

+	/* Find the present coordinates of the window. */
+	XGetWindowAttributes(g_display, g_wnd, &amp;win_attrib);
+	g_last_wnd_pos_x = win_attrib.x + 1;
+	g_last_wnd_pos_y = win_attrib.y + 1;
 	g_fullscreen = !g_fullscreen;

Problem #2:

For some unknown reason, the main display of the guest did not want to resize to the full widescreen resolution of the screen. All the choices were limited to 4:3 resolutions or some weird panoramic setting (no 1680x1050). I couldn't change it through the Windows display properties nor did it auto-resize in any way.

Solution #2:


VBoxManage controlvm name setvideomodehint 1680 1050 32

did the trick. To be sure, I added it to the startup script.

Problem #3:

When using two instances of rdesktop concurrently, at times they have overlapping keyboard focus. I'm not sure why or how, but sometimes after going into fullscreen with one instance it will catch Ctrl+Alt+Enter sent to the other instance (thus disabling fullscreen mode instead of going with two fullscreens). This behaviour is however quite erratic - at times it behaves as it should, and another time it will annoy me for a minute or two.

Solution #3:

I don't have a foolproof solution for this. But it helps sometimes to give the focus to the Linux desktop (or taskbar) first, and then to the second instance of rdesktop. But - as said - it's not always the case. Best way to do this would be to further patch rdesktop with custom key-bindings and have separate hot-keys for both instances, but I didn't have the incentive to do this yet.

U232-P9 USB-Serial adapter drivers


My driver CD for the USB-Serial converter has gone AWOL, so I had to look for the appropriate file. My adapter is a generic “no-name” (PL-2303 based) as can be seen in the photos:

It’s a Prolific chip, so this site should help. I remember not having any problems with the default driver on my desktop (Windows XP). However, this time I had to use the adapter with my netbook (also Windows XP), and it didn’t work with the device manager stating:

This device cannot start. (Code 10)


According to this thread, there’s another version of the driver available. I’m not sure if the one mentioned there is the same as PL2303_Prolific_GPS_1013, though it looks similar. But I’ve tried the GPS version from Prolific, and it did work. My desktop had no problems with the standard version though, so it’s a bit of trial and error method.

Unfortunately, I am unable to maintain this site any longer. The content is available solely for archival purposes and might have become outdated.