JellyBean with multiple users

If you have a rooted Android phone running Jelly Bean, then you can have multiple users, just like on a regular Linux machine. :)
If you want to add a user then open terminal emulator and type:

1. su
2. pm create-user foo

Press and hold the power button and you'll be able to select your newly created user.

If you want to list all users on the system then open terminal emulator and tpe:

1. su
2. pm list-users

If you want to remove a user then open terminal emulator and type:

1. su
2. pm list-users and remember the number before user foo
3. pm remove-user 1

pm list-users gives something like:

Users: 

                UserInfo{0:Primary:3}

                UserInfo{1:foo:3}

Prevent user from running su

If you want to prevent user foo from running command su then do this:

1. sudo grouadd nosu
2. sudo usermod -a -G nosu foo
3. sudo vi /etc/pam.d/su and uncomment
   auth required pam_wheel.so deny group=nosu

Required property "db.example.com" is unknown host. ERRORCODE=-4222, SQLSTATE=08001

If DB2 JCC driver throws mentioned error this means that it cannot resolve db.example.com. You should check your DNS settings.

NetBeans 7.2

NetBeans 7.2 is out!
Go grab it while it's hot. :)
New&Noteworthy page can be found here.

I hope you'll enjoy it as much as I do!

com.ibm.db2.jcc.am.DisconnectNonTransientConnectionException

If you get 

com.ibm.db2.jcc.am.DisconnectNonTransientConnectionException: [jcc][t4][2030][11211][4.13.80] A communication error occurred during operations on the connection's underlying socket, socket input stream, or socket output stream.  Error location: Reply.fill() - insufficient data (-1).  Message: Insufficient data.. ERRORCODE=-4499, SQLSTATE=08001

then this usually means you are using a connection pool to get connections to database and that connections in the connection pool are "stale".

What does stale mean?

You have connection pool started, connections created, but then database restarts. The connection you now get out of the connection pool are now stale and you have to restart the connection pool (close all connections and create new ones).

Sourcing .bashrc in non-interactive shell

This is a new post in a long, long time. A lot has happened since New Year. I'll post about that at a later date. :)

Now to using bash under Linux.
When you are running a bash session in a terminal you are inside of an interactive environment and you can edit .basrc and source it at any given time.
You source .bashrc like this:
source ~/.bashrc or . ~/.bashrc
Now, let's say you have a bash script that you want to run. You can run in in a terminal, but that is not the subject of this post. You can run it via a file manager, like Dolphin, thunar, ... or you can put a shortcut to it on your desktop.
If you want to source .bashrc in that script, it won't work, because running via file manager or desktop is running a bash script in an non-interactive environment. .bashrc notices that and does not do anything. In fact if you look at .bashrc you will find something like this:
# If not running interactively, don't do anything
[ -z "$PS1" ] && return
at the beginning of you .bashrc.
But there is a solution for this. Put PS1='$ ' in your script before sourcing .bashrc.
Examples:
- run a terminal
- edit .bashrc and put 
export TEST_TEXT='My test text' in your .bashrc
- make a new script test.sh:
#!/bin/bash
. ~/.bashrc
echo "TEST_TEXT = $TEST_TEXT"
read
- make it executable
chmod +x test.sh
- run it and you should get something like this:
TEST_TEXT = 
- now source .bashrc and run the test.sh script
. ~/.bashrc
./test.sh
- and you get
TEST_TEXT = My test text
If you run this script via file manager you will always get TEST_TEXT = 
Change the script to 
#!/bin/bash
PS1='$ '
. ~/.bashrc
echo "TEST_TEXT = $TEST_TEXT"
read
and .bashrc will get source everytime you run it.

2012

Let 2012 be a special one.
My New Year's resolution was also to blog more ofter with some interesting dtuff about Linux, version control, Java, ....

Xmarks

Usually I'm using two web broswers: Firefox and Chrome. I always tried having one bookmarks set. I never found a way to share them between those web browsers. I've tried exporting bookmarks from one browser and importing into another, but that didn't feel right. Luckly, yesterday I found Xmarks. Now my bookmarks are in the cloud :)) and I can share them easyly between Firefox and Chrome, since Xmarks has plugins for both.

KDE Konsole

I've been a happy KDE user since the eary days of it.
I've also used Konsole as my primary terminal application, but in recent Kubuntu releases it became slow.
Yesterday I switched to good old xterm with screen:

xterm -bg black -fg grey -fn -*-console-*-*-*-*-*-*-*-*-*-*-*-* -geometry 166x45

and in xterm I start screen with 10 sessions.
Now everything is fast again. :)

Is my application running?

A neat way to check if a instance of an application is already running is to use FileChannel and it's ability to lock it.
An example:

File lockFile = new File("APP_NAME.lock");
FileChannel fileChannel = new RandomAccessFile(lockFile, "rw").getChannel();
FileLock fileLock = fileChannel.lock();

lock() will block, but you could use tryLock() to check it the fileChannel is already locked.
I've been using it in our application on Linux and Windows.

Mercurial server installation for Windows

Installation of hgweb interface:
- install Apache HTTPD 2.2 to c:/Apps/Apache
- install Python 2.6 to c:/Python26
- install Mercurial to C:/Apps/HG
- install EasyInstall
This will find your Python installation and install it there.
- put c:/Python26 and c:/Python26/Scripts in you PATH
- install Flup Python module
easy_install flup-1.0.3.dev_20110111-py2.6.egg
- unzip C:/Apps/HG/library.zip C:/Apps/HG/dev
- copy c:/Apps/HG/templates directory to c:/Apps/HG/dev/templates

I have my Mercurial repositories in d:/HG/repos.

hgweb.fcgi in d:/HG looks like:

 #!C:/Python26/python.exe -u
 #
 # An example FastCGI script for use with flup, edit as necessary
 
 # Path to repo or hgweb config to serve (see 'hg help hgweb')
 config = "d:/HG/hgweb.config"
 
 # Uncomment and adjust if Mercurial is not installed system-wide:
 import sys
 sys.path.insert(0, "C:/Apps/HG/dev")
 
 # Uncomment to send python tracebacks to the browser if an error occurs:
 import cgitb; cgitb.enable()
 
 import os
 os.environ["HGENCODING"] = "UTF-8"
 
 from mercurial import demandimport; demandimport.enable()
 from mercurial.hgweb import hgweb
 from flup.server.fcgi import WSGIServer
 application = hgweb(config)
 WSGIServer(application).run()

- configuration file for hgweb (d:/HG/hgweb.config) looks like:

 [paths]
 # One repository
 #FOO = d:/HG/repos/FOO
 # All repositories in d:/HG/repos
 / = d:/HG/repos/**
 
 [web]
 style = gitweb

- each repository under d:/HG/ has hgrc (in UTF-8) like this:

 [web]
 description = Repository description
 style = gitweb
 allow_push = *
 push_ssl = false
 verbose = false
 allowbz2 = yes
 allowgz = yes
 allowzip = yes
 
 [ui]
 username = Owner of repo <repo.owner@example.com>
 debug = false

NOTE: debug and verbose have to be false

- edit c:/Apps/Apache/conf/httpd.conf and add:

 LoadModule cgi_module modules/mod_cgi.so
 LoadModule env_module modules/mod_env.so
 
 SetEnv FCGI_FORCE_CGI Y
 ScriptAliasMatch        ^/hg(.*)        d:/HG/hgweb.fcgi$1
 <Directory d:/HG>
  Options ExecCGI FollowSymLinks
  AllowOverride None 
  AddHandler cgi-script .cgi
  Allow from all
  AllowOverride All
 </Directory>

- if you want to limit access to repositories add this to c:/Apps/Apache/conf/httpd.conf

 <Location /hg>
  AuthType Basic
  AuthName "Mercurial repositories"
  AuthUserFile D:/HG/hgusers
  Require valid-user
 </Location>

- then add valid users to D:/HG/hgusers like this:

 htpasswd -c D:/HG/hgusers frodo # creates file D:/HG/hgusers and adds user frodo
 htpasswd D:/HG/hgusers sam # only add usersam to exising D:/HG/hgusers file

DB2 SQL440 or SQL901 on DELETE

If you get SQL440 or SQL901 when you try to delete from a table, you might encountered this problem. The solution is to recreate all foreign keys that point to this table (MY_TABLE is the SQL script below).
The SQL to generate that SQL script for you is this (statement delimiter is #):

SELECT 'ALTER TABLE EMGSYS.' || TABNAME || ' DROP CONSTRAINT ' || CONSTNAME || '#' || CHR(13) || CHR(10) ||
'ALTER TABLE EMGSYS.' || TABNAME || ' ADD CONSTRAINT ' || CONSTNAME || ' FOREIGN KEY (' || TRIM(FK_COLNAMES) || ') REFERENCES EMGSYS.' || REFTABNAME || '(' || TRIM(PK_COL
NAMES) || ') ON DELETE ' ||
(CASE WHEN DELETERULE = 'R' THEN 'RESTRICT' WHEN DELETERULE = 'C' THEN 'CASCADE' END) || ' ON UPDATE ' || (CASE WHEN UPDATERULE = 'R' THEN 'RESTRICT' WHEN UPDATERULE = 'A
' THEN 'NO ACTION' END) || '#'
FROM SYSCAT.REFERENCES WHERE REFTABNAME = 'MY_TABLE' ORDER BY TABNAME WITH UR

There are still some constands missing in DELETERULE and UPDATERULE. The various values can be seen here.

Mercurial

At our company we used (well, still using) CVS as our versioning system. We've finally moved one of our main projects to Mercurial. I've been given a task to expose the repository over HTTP and securing access to it. In the following days I will post how I did that on Windows. Oh, well, I know. I'd rather use Linux, but what can you do....

New Year

I wish to you all a better 2011

Is this for real?

Read what Bjarne Stroustrup said about creating C++ here. Then tell me what you think.
I must admit I always found C++ a bit weird. :)

More space on Linux HOME partition

I've been trying to post this for a while, but never came to it.
Here it goes.

On every Linux partition there is a certain percent (usually 5%).
For certain partition you can check the amount of reserved space like this:
sudo dumpe2fs /dev/sda3 | grep "Reserved block count"

On my computer I had about 6Gb of reserved space.
You can gain those 6Gb of space back simply by using:
sudo tune2fs -m 0 /dev/sda3

I DON'T recomend using this on you root, boot, ... partitions. More on this can be read at How to Gain Couple Of Gb Of Free Space On Linux

First

Two days ago I bought an HTC Desire phone with Android 2.2. I'm still learning, installing applications, ... This is my first smart phone and it's great. :))

This is also my first post from the phone.

Compile DB2 plugin for Qt on Linux

Steps:
- get Qt from Nokia's page
- install it
- run configure like this:

./configure -plugin-sql-db2 -v -I /opt/ibm/db2/V9.7/include -L /opt/ibm/db2/V9.7/lib64/
make
make install (if you want to install it)

- this should compile DB2 plugin for Qt. Now you can use this installation or copy generated Qt db2 plugin (libsqldb2.so) to /usr/lib/qt4/plugins/sqldrivers

Access DB2 via ODBC on Linux

This is how to install ODBC on Linux (Kubuntu 10.04) and access it.
Steps:
- install DB2 (I always install everything)
- install unixODBC
- edit $HOME/.odbc.ini :

[DB_NAME]
Description = Connection to DB_NAME
Driver=$HOME/sqllib/lib/libdb2.so

- run db2ca and add DB_NAME (don't forget to add ODBC)
- now you can use it on Qt for example like this:

QSqlDatabase db = QSqlDatabase::addDatabase("QODBC");
db.setDatabaseName("DB_NAME");
db.setUserName("DB_USER");
db.setPassword("DB_PASS");
db.open();
QSqlError error = db.lastError();
qDebug() << "db error = " << error;

QSqlQuery query = db.exec("SELECT 1 FROM SYSIBM.SYSDUMMY1");
while (query.next() == true) {
qDebug() << query.value(0).toString();
}

DB2 9.7.1, db2start and SQL1042C

As you may know I'm running Kubuntu 9.10.
The other day I installed DB2 Express-C 9.7.1 on it. When I tried db2start I got:
SQL1042C An unexpected system error occurred. SQLSTATE=58004

If you get the same try running db2ftok and then db2start again. It works for me.