Bits By The Pound

After upgrading to Ubuntu 9.10 from 9.04, the GWT plugin for Eclipse doesn’t work so well when trying to run an app in hosted mode.

** Unable to load Mozilla for hosted mode **
java.lang.UnsatisfiedLinkError: /home/igough/installs/eclipse-3.5/plugins/com.google.gwt.eclipse.sdkbundle.linux_1.7.1.v200909221731/gwt-linux-1.7.1/mozilla-1.7.12/libxpcom.so: libstdc++.so.5: cannot open shared object file: No such file or directory

Karmic does not have libstdc++.so.5 as part of the distribution but you can download the one from 9.04 from the Ubuntu repository. I am running 64 bit Ubuntu, and would normally download the 64 bit version of the library, but the GWT Eclipse plugin uses a 32 bit version of Mozilla for hosted mode. And its hosted mode that wants the Standard C++ library. So we must download and install the 32 bit version.

Download libstdc++5_3.3.6-17ubuntu1_i386.deb from the Ubuntu respository link above. Then install using:

Once you do this, hosted mode should work again.

Today, I upgraded my laptop, currently running Ubuntu 9.04 to Ubuntu 9.10.

In general, 9.04 has seemed to have been an ill-behaved release with all sorts of problems handling the Intel GM965 chipset in my Inspiron 1720, in particular the X3100 graphics subsystem. It got so bad for JP that he moved his laptop over to a very pre-release version of Karmic a few months back.

One of the new features that runs on Karmic is that Palimsest Disk Utility 2.28 which you can find at System->Administration->Disk Utility. This utility monitors the SMART data from all drives in your system to let you know when the disks start getting flakey. When JP upgraded, the utility warned him that his disk was starting to go bad, with dire warnings about the disk failing at any moment. JP replaced the stock Fujitsu drive with a Seagate 500Gb drive and all has been well. When I did the Karmic upgrade, Palimsest gave me a disk failure warning too. I looked at the SMART data that the utility was reporting and marveled that the numbers that were being reported were so high. So then I ran Spinrite 6 against the drive. Spinrite did not detect any problems, but did show the same oddly high SMART numbers. So I ran smartctl

> sudo smartctl -a /dev/sda

Now this pointed out that the particular drive, a FUJITSU MHY2160BH is not in the smartctl’s database. Apparently, SMART isn’t as much a standard as it should be. Different vendors do different things with the SMART attributes. My suspicion is that if the particular drive is not in the database, the program cannot know how to interpret the raw numbers correctly. Palmisest and smartctl are show the same raw values for the SMART attributes.

Lets take a look at some of the raw numbers:

Reallocated_Sector_Ct   8589934592000

Wow, thats a lot of reallocated sectors. 8 trillion. On a 160Gb hard drive. I’m not buying it. But that number in hex is 07D0 0000 0000, an oddly round number. Maybe its not a 48 bit number in this case, but rather a 16 or 32 bit number and the real value is zero.

Power_On_Hours  897474

Lot of hours. I know the drive has not been powered on for 101 years, so this number is suspect too. Converting to hex gives 000D B1C2, but I cannot relate any portion of this number to reality. I would expect in the order of 16800 hours (0x41A0).

Ok, so now I don’t trust anything that Palimpsest is telling me. Maybe its wrong. Spinrite thinks the drive is fine and that is good enough for me. I only have the single drive in my laptop and have automatic daily backups and regularly run Spinrite on the drive. So if Palimpsest is not telling me anything useful, no sense in it taking up memory, so I have stopped it from starting when my laptop boots.

Today, JP and I built our first Google Wave Robot. It was quite an excercise in frustration. We followed this tutorial and found that it is well put together and straightforward.

After creating an App Engine account, we built the tutorial as-is in order to verify the procedure was correct. Simple enough until we tested the Robot. The code looks like it was supposed to emit “I’m Alive!” whenever it was added to the Wave and “Hi, Everybody!” when any other participant was added. But we ended up seeing “I’m Alive!I’m Alive!” in one Wavelet and “Hi, Everybody!Hi, Everybody!” in another Wavelet. Examining the code and examining the wavelet data in the debug view we could not figure out any way that using TextView.append would cause this unless something was horribly broken. For example, the following code:

would output the text twice in the same blip.

So we started printing out all sorts of information in order to figure out what was going on. At this time, we discovered that putting a newline (\n) into the blip did not actually cause a line break. Well, of course. It must be the HTML <br> element. Nope. Nothing we tried worked. But this was just a small annoyance, making the returned debugging text difficult to read.

After hours of testing, we stumbled upon a post pointing out  the bug for the doubled text. A nice fresh bug, 2 days old. Causes repeated text when using the TextView.append() method. At least there is a workaround. If you do a delete() on the TextView prior to the append(), the text does not get repeated.

Then we discover a post that shows that the newline used to work, but now you need to use spaces around the newline for a linebreak to occur. eg. ” \n “. Note the spaces below:

Now that we have the sample working the way it should, we can modify it to accomplish something useful. Hopefully the process will not be as frustrating.

Both JP and I have experienced hard drive failures on the laptop just outside of the warrenty period. The new drives (not from Dell) have been working fine and require less power.

To get a temperature display of your laptop.
> sudo apt-get install sensors-applet

Right click on the top dock area and select “Add to Panel…”
Scroll down to the Hardware section and click on Hardware Sensor Monitor.

To add a second drive, you will need:

Dell parts

C7586 Hard Drive Carrier for 2nd HDD bay
XK231 Interposer (interfaces HDD to laptop)
The place below has the parts, but they are not Dell parts, hence cheaper.
http://newmodeus.com/shop/index.php?main_page=product_info&products_id=144

latest Ubuntu update for compiz says it is un-blacklisting the i965
from compiz because there is a better workaround in the intel driver.

If you have already done the upgrade as described in
http://ubuntuforums.org/showthread.php?t=1130582 then you will need to
revert your changes to pick up the fixed intel driver.

I have installed it and it seems to fix the hanging problems that I
was seeing prior to upgrading as in the link above. Unfortunately, my
touchpad is just as insensitive as before. It drives me nuts.

http://ubuntuforums.org/showthread.php?t=1130582

tried this and the Graphics worked a lot better, but the touchpad started working extremely poorly. Had to revert the changes.

Experiencing problems in Jaunty with hanging on reboot and shutdown.

http://ubuntuforums.org/showpost.php?p=7149271&postcount=22