Selected Unix/Linux tips:

To catch standard error output to a file:

• Add 2> <filename> to the command line.
• Note: 1> <filename> is the same as ><filename> and pipes standard output to <filename>
• The standard error output is often used rather than standard output by programs to put messages to the screen.  SAM puts a particularly voluminous message stream to standard error, which is too long to scroll back through.  This maneuver saves the log of messages into a file for future examination. The messages will not come to the screen, however.
• 2>&1 <filename> captures both standard output and standard error to a single file.

• Open a new terminal window.
• Or else in the same window, put the program in the background by <control>Z followed by bg.
• Or else start the program in the background by ending the command line with an &.
• type tail -f <filename>
• The end of the file will be displayed and updated each time it is written to.
• You will have to terminate the tail function with <control>C when the output is over.  Therefore if you use the background method, you will have to be able to recognize the end of the output.

• Suspend it with <control> Z
• Type ps to get the job number.
• type kill -15 <job number>
• This tries to close output files and kill the job.
• If this fails, type kill -9 <job number>
• This kills without trying to save files.
• If the job has been put in the background with bg,  then bring it to the foreground with fg, then kill as above.
• If the job was put into the background and then you logged out, you can log back in and kill it by:
• ps -u <your username>
• Find the job number from the list and kill as above.
• Note: if you try to kill someone else's process by mistake, you will be denied permission.

• Put the following file by the name gcg.com in your home directory


source /home/users/gcg/gcgstartup.ksh
gcg
set +o noglob
 

• At startup and any time you experience this problem, type source ~/gcg.com

• The gcg startup file executed at startup for most bioinf users inactivates wildcard usage.  To reverse this, type set +o noglob.
• As indicated in the item above, you can put this fix in a gcg initialization file.
• Note: ls * lists the directory and all of its subdirectories; just ls lists the whole directory;  and ls *.* lists all files with extensions in just the default directory. ls */ -d lists just the names of subdirectories in the directory. ls -I*.* list just entries without an extension.

• To create a symbolic directory link:  ln -s {path to existing directory} <symbolic directory name>
• This causes the symbolic name to appear in the default directory followed by an @.  Addressing the symbolic name as if it were a directory name will be the same as addressing the existing directory to which it is linked.
• You can create your own symbolic links to reduce the amount of typing in commands.  You can remove them with the rm command.
• If you follow a symbolic link with cd, the prompt and the pwd command will reflect the symbolic name and path of the directory rather than the hard name and path.  To find the hard name, go to the directory with the symbolic link and use ls -l.

• Type locate <filename>
• Will find all files in system containing <filename> as a part of the filename.
• This requires that the sys adminiatrator is keeping a database of all files in the system up to date in a default location for the locate program.

• If you want to retain output that would come to the screen, start the job with >out.log 2>error.log
• These two files (which could be any names you like) will retain the standard output and error output that would otherwise come to the screen.
• After starting the job, type <control Z> to interrupt it.
• Then type bg to restart it in the background.
• You can then do other things in the terminal window, including logging out.  The job will run until finished.
• You can examine the two log files or other files written by the program to observe the progress of the program, using the cat or tail commands to type the contents of a whole file or just the end of it, respectively.
• In order to tell if the job is finished when you log back in, you will need to examine the output files and be able to identify an indication of completion.

• Before starting job type screen, which starts a new shell.
• Start job in that shell.
• Put the whole shell in the background with <control A> D
• After logging out and logging in later, find the screen session by typing screen -list,
• You'll get something like this:


           There is a screen on:
                   17279.pts-0.triton      (Detached)
           1 Socket in /root/.screen.
 

• Then reattach to that screen by typing screen -r 17279
• Alternative: If you start the session through vncviewer in the first place, then you don't need to do any of this.  Just close the vncviewer window without logging off kde.  When you connect again from elsewhere with vncviewer, you will rejoin the same session in progress.

• Detailed instructions are found by typing man tar
• The following examples found in some tar man entires but not others are helpful:
• EXAMPLES

       tar -xvvf foo.tar
              extract foo.tar

       tar -xvvzf foo.tar.gz
              extract gzipped foo.tar.gz

       tar -cvvf foo.tar foo/
              tar contents of folder foo in foo.tar