OK, flat and messy is better for databases.  If the data exists already get it in a database first and then add structure that you need later.

Don’t waste time on what you never need.

I have some old-ish java software that I needed to get going on Fedora Core 8. The icedtea java stuff never seems to work for me. I usually pull down the java sdk from Sun and then manually change all the things in the path, mozilla directories and my environment variables until I can get java to work. I decide this time to invest some time trying to figure out the right way to do it.
First I installed java (1.6 sdk), Sun now puts their stuff in /usr/java. There is a /usr/java/default link to a latest directory which links to the version I installed.

[root@h ~]# ls -l /usr/java
total 16
lrwxrwxrwx 1 root root   16 2008-08-07 13:18 default -> /usr/java/latest
drwxr-xr-x 9 root root 4096 2008-08-07 13:18 jdk1.6.0_07
lrwxrwxrwx 1 root root   21 2008-08-07 13:18 latest -> /usr/java/jdk1.6.0_07

FC8 has a utility called alternatives. This utility controls the links to standard utilities so that your machine can have different packages of these utilities and you can select which package you use. For example, java (the run time VM):

lrwxrwxrwx 1 root root 22 2008-08-01 09:44 /usr/bin/java -> /etc/alternatives/java
lrwxrwxrwx 1 root root 26 2008-08-18 16:30 /etc/alternatives/java -> /usr/java/default/bin/java

The alternatives utility set this link up. Alternatives has a concept of a master links and slave links. Master links have a name (e.g. java). The master’s slave links follow the master link. When you select a different link with alternatives, it changes the master link and the slave links.

To see the different options for a master link, you can run ‘alternatives –display <name>’. For example:

]# alternatives --display java
java - status is manual.
 link currently points to /usr/java/default/bin/java
/usr/lib/jvm/jre-1.5.0-gcj/bin/java - priority 1500
 slave keytool: /usr/lib/jvm/jre-1.5.0-gcj/bin/keytool
 slave rmiregistry: /usr/lib/jvm/jre-1.5.0-gcj/bin/rmiregistry
 slave jre_exports: /usr/lib/jvm-exports/jre-1.5.0-gcj
 slave jre: /usr/lib/jvm/jre-1.5.0-gcj
/usr/java/default/bin/java - priority 2000
 slave keytool: /usr/java/jdk1.6.0_07/bin/keytool
 slave rmiregistry: /usr/java/jdk1.6.0_07/bin/rmiregistry
 slave jre_exports: /usr/lib/jvm-exports/jre-1.5.0-gcj
 slave jre: /usr/java/default/jre
Current `best' version is /usr/java/default/bin/java.

The output here says that the java link is set manually. (Automatic would set it best on the priority). The link currently points to /usr/java/default/bin/java. You can also see the slave links that go along with it.

To change the link, you can run ‘alternatives –config java’.

]# alternatives --display java
java - status is manual.
 link currently points to /usr/java/default/bin/java
/usr/lib/jvm/jre-1.5.0-gcj/bin/java - priority 1500
 slave keytool: /usr/lib/jvm/jre-1.5.0-gcj/bin/keytool
 slave rmiregistry: /usr/lib/jvm/jre-1.5.0-gcj/bin/rmiregistry
 slave jre_exports: /usr/lib/jvm-exports/jre-1.5.0-gcj
 slave jre: /usr/lib/jvm/jre-1.5.0-gcj
/usr/java/default/bin/java - priority 2000
 slave keytool: /usr/java/jdk1.6.0_07/bin/keytool
 slave rmiregistry: /usr/java/jdk1.6.0_07/bin/rmiregistry
 slave jre_exports: /usr/lib/jvm-exports/jre-1.5.0-gcj
 slave jre: /usr/java/default/jre
Current `best' version is /usr/java/default/bin/java.
[root@fstats ~]# alternatives --config java

There are 2 programs which provide 'java'.

  Selection    Command
-----------------------------------------------
   1           /usr/lib/jvm/jre-1.5.0-gcj/bin/java
*+ 2           /usr/java/default/bin/java

Enter to keep the current selection[+], or type selection number:

Then you just pick the number you want. The problem now is how do you have Sun’s java as an alternative? After playing around for a bit, here’s the script that worked for me:

#!/bin/bash

alternatives --verbose --install /usr/bin/java java /usr/java/default/bin/java 2000 \
  --slave /usr/lib/jvm/jre jre /usr/java/default/jre \
  --slave /usr/lib/jvm-exports/jre jre_exports /usr/lib/jvm-exports/jre-1.5.0-gcj \
  --slave /usr/bin/keytool keytool /usr/java/jdk1.6.0_07/bin/keytool \
  --slave /usr/bin/rmiregistry rmiregistry /usr/java/jdk1.6.0_07/bin/rmiregistry

alternatives --verbose --install /usr/bin/javac javac /usr/java/default/bin/javac \
2000  --slave /usr/lib/jvm/java java_sdk /usr/java/default \
  --slave /usr/lib/jvm-exports/java java_sdk_exports /usr/lib/jvm-exports/java-1.5.0-gcj \
  --slave /usr/bin/javadoc javadoc /usr/java/default/bin/javadoc \
  --slave /usr/bin/javah javah /usr/java/default/bin/javah \
  --slave /usr/bin/jar jar /usr/java/default/bin/jar \
  --slave /usr/bin/jarsigner jarsigner /usr/java/default/bin/jarsigner \
  --slave /usr/bin/appletviewer appletviewer /usr/java/default/bin/appletviewer \
  --slave /usr/bin/rmic rmic /usr/java/default/bin/rmic

There is stuff out on the web that kind of explains this but I had trouble with a few things. Here is the man entry for the config command:

alternatives [options] --install link name path priority [--slave link name path]... [--initscript service]

The link is the link on your path (e.g. /usr/bin/java) not the link in the /etc/alternatives directory. The name is the master link name (e.g. java). The path is where you new stuff is (e.g. /usr/java/default/bin/java). The slave links have names too.

Terrific, how do you find out what all nasty slave links and names should be in your ‘alternatives –config’ command line? You can look at ‘alternatives –display’ output but it doesn’t tell you install link (e.g. /usr/bin/java). I don’t know why this is — you should be able to use the tool to display everything you need to install an alternative.

Anyway, here’s the punch line. You can goto the /var/lib/alternatives directory and there is a file in there for every master link. For example, here is my /var/lib/alternatives/java:

auto
/usr/bin/java
keytool
/usr/bin/keytool
orbd
/usr/bin/orbd
pack200
/usr/bin/pack200
policytool
/usr/bin/policytool
rmid
/usr/bin/rmid
rmiregistry
/usr/bin/rmiregistry
servertool
/usr/bin/servertool
tnameserv
/usr/bin/tnameserv
unpack200
/usr/bin/unpack200
jre_exports
/usr/lib/jvm-exports/jre
jre
/usr/lib/jvm/jre
java.1.gz
/usr/share/man/man1/java.1.gz
keytool.1.gz
/usr/share/man/man1/keytool.1.gz
...

Note that there are a bunch of entries that aren’t listed by ‘alternatives –config’ or ‘alternatives –display’.  I ignored those when I wrote my ‘alternatives –install’ script.

I am sure that there are things that I did wrong here but it seems to work and allows me to switch back and forth between Icedtea and Sun’s java.

Spent a lot of time this morning trying to get a Creative Webcam on Fedora Core 8 (FC8). My goal was to get streaming working from the webcam. Man, this was a pain and its still not really working. Here is what I did.

I plugged the webcam in. It didn’t just work — nothing popped up on my desktop. Dang.

[root@bigbird ~]# lsusb
...
Bus 005 Device 002: ID 041e:4034 Creative Technology, Ltd
...

Then I dorked around on the net and starting yum-ming everything that I could find related to webcam.

The one that seemed to work is here. I kinda half followed the directions. The atrpms was already in my /etc/yum.repos.d/atrpms.repo file. I just enabled the stable by setting enabled=1 (it was 0). I then did the following.

yum install gspcav1
yum install uvc
yum install camE
yum install camstream

I rebooted and camE worked.

I then tried to figure out streaming. Man, what a confusing mess. I pulled up vlc. When you open the device in vlc (video0), at the bottom of the page is a box for streaming. Select that box and then click on settings. Select “play locally” and “HTTP”. For the encapsulation method select “Ogg”. For the transcoding options, select “Video Codec” and then mp4v in the drop down. Under Miscellaneous check “Select all elementary streams” and set the time to live to 8. Then press OK.

To view this stream with mplayer, try the following:

mplayer http://localhost:1234

The ogg encapsulation is the only encap that seemed to let mplayer find the fps. I could not figure out how to find the fps on the other methods.

Note that I believe that you will need to fix the firewall (iptables) and probably something in SE Linux to get this stream off your linux box. Here’s a link that might help.

I was in a meeting discussing minimal security for a home network application. The question on the table was whether the family users need or want separate identities when interacting with network devices.

Thinking about it, it seems like taking the computer security model (e.g. you log onto to your work station) and apply it to generic devices that are on the network seems silly. For example, do you log into the refrigerator or the toilet if it happens to be networked (a.k.a. living in Hong Kong)? This seems stupid but these devices monitor the user’s health. How can they do that without the current user being identified?

From 30 years of computer security research and even more of non-computer security work, identity is required for security. But I think this identity is really only a tag for a set of privileges and does not necessarily need to be a single individual. I do think that for security of large group of people you need to identify each person (or more accurately each login should be given to only one person) but I believe for most families this is not necessary in general and in general identity is more a feature enabler (i.e. your toilet tracking your individual hydration and telling you to drink more water) than a set of privileges. The set privileges (e.g. watching a rated R movie) can be accessed by people who prove they are in the adult group by knowing the adult password.

To that end here is patent that seems to use the remote control to identify the TV viewer: link

Here’s a scan of the specs of a computer that I bought in 1989. At this time, I was a Junior Electrical Engineering major at A&M. We had VAX accounts at school that we used to run SPICE models and email. In the few CS classes that I took, we had UNIX accounts. The 386SX was used for remote access via A&M’s dial-up (terminal), word processing, and Quatro-Pro for some lab results.

Look at how much it cost me!!!!

A few years later, I bought a 486DX-33 computer. I was still at A&M but working on a Master of Science degree. I used it and turbo C++ and gcc based compiler on DOS to run simulations of Speech CODECs and packet loss recovery techniques. I also ran Latex on it to write my thesis.