focus behavior refers to the condition that determines which window recieves input from the keyboard. The standard rule is ``click to focus'' which means that the window on which you last clicked holds the focus until you click somewhere else.
It will save litterally minutes (over a your life time) if you select instead what is called sloppy of focus follows mouse under which behavior, the window in which the mouse currently resides gets the focus.
This allows you type into partially obscured windows and generally works better with X-style select and past functions.
Under Gnome, the focus behavior is set via the
[System]
[Preferences][Windows] menu.
[System][Preferences][Keyboard shortcuts]
is for doing this. Click on the function of interest and then, when prompted, type a keyboard combination to asign to that function.
The ``aplets'' that live on your various ``panels'' can be easily
created and accidently destroyed. The two most useful applets are the
Workspace Switcher applet which shows your ``virtual
desktops'' and allows you to switch between them;
and the
window list applet which shows your running applications.
To (re)create these or other panel applets,
on the panel to which you wish to add an applet. Then scroll to
select the applet of interest. Window List and Workplace
Switcher are near the bottom. Click the
to add the applet.
There are lots and
lots of gizmos and gimcracks besides Workspace Switcher and
window list that you can add this way.
If the application in question is in a menu, you can simply drag and
drop the thing from the menu to a panel. Just select the item in the
menu; click and hold
; then drag it to the panel wherein it
belongs.
If the item is not in a menu - or if you just like doing things the
hard way, you can add a ``launcher'' by clicking
over the panel to
which you wish to add the launcher and selecting
[PanelAdd to panellauncher].
The resulting dialog box allows you to specify the command that launches the program and whether or not it should be run in a terminal window.
If the title bar is not obscured then just click and hold
the
over the title bar. If the title bar is obscured then of course you
cannot do the this. In this case your best option is to make sure that
the window you wish to move has the focus, then hit
+
. Click the
to release the window in a new location.
You can also move a window by clicking and dragging the
+
on its representation in the virtual desktop and dragging it.
The desktop in the current release of Fedora has a panel at the top of
the screen. If you need to recreate it
click the
on an existing panel, then select [New
Panelmenu panel]. The new panel might appear along a side
instead of the top, if it does, then click
and select
[properties] to change its ``orientation'' to something to your liking.
If the file is ASCII, PostScript, or pdf, you can simply type
@:
a2ps -2 -g -P printer filename
The -g flag causes a fancy border to be drawn around each page the -2 flag indicates that 2 pages per sheet is sufficient
For those who cannot be bothered with typing commands, there is also a GUI alternative, see 2.2.
xpp is your best option. xpp offers you menus and buttons that allow you to change the printer features. For example you can print n-up, or choose toner saving modes and perhaps do other morally virtuous printer related things.
To launch type:
@:
xpp filename
Add paper - there boxes of reams in the Xerox room. Why not bring as many reams as you can carry without endangering your health and save other's the trouble.
Install a new toner cartridge - or if you're too chicken at least tell a grown up that something is wrong. Toner cartridges live in the Xerox room. They should be labeled as to which printers the will work in - if not check with Monique. Either way send Monique email informing her that you have used a toner cartridge so that another can be ordered.
Instructions for setting up your computer to print to demography
printers when connected via AirBears is at
http://lab.demog.berkeley.edu Under [Documentation]
[Printing from Portables].
Several data file formats can be imported directly in to R via the foreign library. If the sort of data you want to import is not on the list below, then there is still a good chance that you will be able to use stat-transfer to convert it to one that is recognized by R.
Stat-transfer can convert just about any kind of file into a SAS transport format file or an ASCII delimited file, either of which can be read by R. Some information - such as SAS informats and labels - will be lost in this process, as there are no equivalent structures in R.
To read ASCII delimited files, the easiest method is read.table()
To read data from a tab delimitted ASCII file called 'truth.csv' with the first row of the file containing variable names the following should suffice:
;
Note the following gotchas with read.table():
To read and write non-ASCII types of files, use the foreign library. To access it you must enter:
To see how it works type:
Currently, the the foreign library supports the following file types
S3 read functions Read an S3 Binary File
read.dta Read Stata binary files
read.epiinfo Read Epi Info data files
read.mtp Read a Minitab Portable Worksheet
read.spss Read an SPSS data file
read.ssd obtain a data frame from a SAS permanent
dataset, via read.xport
read.xport Read a SAS XPORT format library
write.dta Write files in Stata binary format
All of these functions expect to be passed a file parameter as in:
![\fbox{\parbox[c]{.8\textwidth}{Be patient, \texttt{foreign} library functions ca...
...utes to be converted and read in. Unless you get an error
message, get coffee}}](img13.png)
Like Stata, R insists on holding all objects in core memory, this makes it a poor choice for data sets that are larger than 500 MB. Since tapinos has 10GB of RAM, large datasets can be processed in R and Stata, but you will surely want to do such things in batch mode.
@:
st
Import your SAS transport-format file into R:
objectname will be a dataframe.
Some files can be imported directly using methods contained in the foreign library, these include SPSS, Stata, Epiinfo and minitab (see the help pages on packageforeign for details). For other types, you may need to convert the file into a SAS transport or other convenient type. Stat-transfer (See Section 4.4.2) is great for this. Once your file is in say SAS transport format (See Section 3.1.1 for instructions on importing it).
Below is a chunk of R code that will draw a pretty nice population pyramid, please adapt it and improve it.
########################################################################
## Sat Mar 9 20:21:19 PST 2002
## Carlm's pretty good population pyramid function.
########################################################################
poppyr<-function(age.male,age.female,agecats=c(0,1,4,seq(5,100,5))){
##age.male is a vector of ages of males, age.females is a vector of the
## mean batting averages of the New York Yankees, agecats is a
## vector of break points of ages -- defaults to a pretty standard
## one
##
## Will draw a pretty good but not quite perfect population pyramid
male.ac<-table(cut(age.male,breaks=agecats))
female.ac<-table(cut(age.female,breaks=agecats))
## 5/13/04 Megan Heller reports that some alternative 'figs'
## values can improve the appearance of the population pyramid
## by eliminating the space between the sexes. Here are
## Megan's suggested values:
# ##this one looks best in terminal
# split.screen(figs=rbind(c(0,.58,0,1),c(.43,1,0,1)))
## this one looks best in png (png files are the easiest to
# import into MSWord -- not that you'd ever want to do that.
# split.screen(figs=rbind(c(0,.58,0,1),c(.38,1,0,1)))
split.screen(figs=rbind(c(0,.57,0,1),c(.45,1,0,1)))
screen(1)
barplot(female.ac,horiz=T,names=paste(agecats[-length(agecats)]),
xlim=c(max(female.ac)*1.1,0),col="#FF9900")
title("Female")
screen(2)
barplot(male.ac,horiz=T,axisnames=F,
xlim=c(0,max(male.ac)*1.1),col="#0000FF")
title("Male")
close.screen(all=T)
}
age.male<-rnorm(10000)*50
age.female<-rnorm(10000)*50
poppyr(age.male,age.female)
mtext(side=3,line=3,text="Population Pyramid",cex=1.5)
You might want to do this if you have two vectors of data which have very different scales but which are both thought to be related to a third vector. In other words, you have two y-vectors which you would like to plot against a single x-vector.
The trick is to use the par(new=T) command, which tells R not to reinitialize the plotting device with subsequent high level commands (such as plot() which would ordinarily cause this to happen. Instead, R will superimpose subsequent plots on top of what is already displayed on the current device.
You will most likely also want to use axes=F argument to one of the plot() commands. This argument supresses the drawing of axes.
##################################################
## this will create a scatter plot displaying
## two sets of points, one corresponding to the
## y axis on the left and one to the y axis on the
## right
##################################################
dev.off() ## start with a new graphics device
# X11() or postscript()
plot(x<-rnorm(100),y<-rnorm(100))
z<-rnorm(100)*250
par(new=T) ## Tell R not to reinitialize graphic device
## for subsequent plots
plot(x,z,col='blue',axes=F)
axis(side=4,col.axis='blue')
par(new=F)
![\fbox{\parbox[c]{.8\textwidth}{NOTE: in the example, the two y-vectors are plott...
...cal, you can show the second x-axis by
typing: \texttt{axis(side=1,line=1)}
}}](img16.png)
(Thanks to tmiller) There should be an easy way. Meanwhile, here is a hard way that works:
# Set margins on yaxis side large enough (10.1?) to accomodate ``foo'' par(mar=c(5.1,10.1,4.1,2.1)) par(las=1) # print axis numbers horizontally # Don't print y-axis text label plot (x,y, ylab=" ) # Print y-axis text label horizontally mtext(``foo'',side=2,line=4)
Load the library:
Create an object of type ``xtable''.
Print the object to a file:
Inside your latex file add:
\input{table42}
The xtable() function has special tricks for several objects that it knows about. For example:
To see a list of objects that xtable() knows about type:
See Section 3.3.1 for a description of how to make latex tables from R objects. the xtable() function can be instructed to create HTML instead of latex. Also see Section 7.0.4 for a description of how to translate latex into very nice HTML.
Files such as jpeg, gif, bmp, pnm, png, tiff, ppm and lots of others. What distinguishes these file formats from say PostScript, is that the information they contain is where to put a bunch of dots. PostScript, WMF, CGM and others are ``Vector'' formats. They contain information on lines and shapes which the computer then converts into points when rendering it on the screen.
ps2pdf13 runs on the linux machines. There are 3 versions of ps2pdf. Ps2pdf14 produces output compatible with Acrobat 4 and later. If you need something compatible with more ancient versions check the man page for ps2pdf.
@:
ps2pdf14 filename.ps
Each of the above commands produces a file called filename.pdf. NOTE: if your postscript file is the output of dvips it will look much better if dvips was given the -Pcms option. BUT more to the point, you can convert .dvi to .pdf directly. See( 7.0.3).
On a linux machine, this should work:
@:
cat first.ps second.ps third.ps | ps2pdf14 - all.pdf
@:
gs -sDEVICE=ppm -sOutputFile=filename.ppm filenmame.ps(or pdf)
@: xv filename.ppm
@:
display filename.ps
@:
gimp filename.ps
Both gimp and ImageMagic can be used to edit and convert among graphic types also.
Generally this is useful for including a postscript file in MSWord.
Use epstool:
@:
epstool -t4 -ofilename.eps filename.ps
Then in Word: [Insert][Picture][From File].
The low resolution .tiff preview will appear when editing in Word, the
high resolution postscript file will appear in the printout.
@:
dvips -Pcms -ofilename.ps filename.dvi
pdflatex filename
where filename is the .tex version of the file.
The documentation can be found at http://netpbm.sourceforge.net/docVersion 10-6 is installed for the linux workstations.
Note that netpbm does not have a graphic user interface. These are utility programs that work from the command line. If you want to edit graphics files see 4.3.2
There are lots of choices for editing graphics files. The most comprehensive and therefore complicated is ``gnome image manipulation program'' or gimp. Gimp can edit just about any file format including postscript and it can do just about anything to it. It is not easy to use but it is very cool.
To launch gimp either find it in the menu or type
@: gimp
Lots of documentation is available at http://www.gimp.org/tutorials.html.
A somewhat simpler graphics editing program is ImageMagic to launch it type
@:
display filename
documentation is available at http://www.imagemagick.org.
The hexbin program (on linux) will convert macintosh files to something readable. Use the -3 flag and look for the some-file-name.data.
@:
hexbin -3 filename
NOTE: some-file-name need bear no relation whatsoever to filename. hexbin -3 will give its decoded file(s) names which are encoded in the binhex file itself. The binhex file can be named anything. You probably got it as an email attachment and saved it under some arbitrary name. For this reason it might be smart to unpack your binhex files in a nearly empty directory.
To launch stat-transfer type:
@:
st
then from the stat-transfer prompt () you can generally convert
files using the cp as
cp
filename.ext1 filename.ext2
where ext1 and ext2 represent the more or less standard file suffixes. To see which ``standard'' file suffixes mean what to the current version of stat-transfer type:
?formats
http://lab.demog.berkeley.edu/Refs/stat-trans.unixman.pdffor more detailed instructions.
NOTE if you happen to need to convert files that are over 2GB in size, st will probably fail. A special version of st, called stl exists on our system. It is not a production version, so it probably will fail in ways that st will not, but for large files it's worth a try.
@:
antiword filename | less
@:
antiword -p letter filename | lpr
There are currently two scanners one is Most likely in the basement lab connected to census the other is probably in the library connected to logit. The scanner in the basement is a ``document'' with a 20 page ``auto document feeder''. The scanner in the attic is a ``flat bed'' scanner. The forer is suitable for scanning one or two sided documents (probably into .pdf format) that latter is useful for scanning images one at a time.
So, once you have a shell on census just type:
@:
scan2pdf -o document.pdf
where document.pdf is the name of the pdf file that you would like to create from the scanned images. scan2pdf will prompt you as to whether you like 1 or 2 sided scanning.
![\fbox{\parbox[c]{.8\textwidth}{When you're finished, why not turn off the scanner and thereby
save the world.}}](img23.png)
Before going down this road ask yourself it it is posible to either convert a .pdf into what you want or to use .pdf instead. It's going to be a lot simpler if you can. See Section 4 on file format conversion for more information.
Well if you must scan into something other that .pdf, then you'll need to use the scanadf command. The scanadf command produces a file for each page in your stack of documents. Those files will be in .pnm format, which though seldom used, is easily converted into just about anything via a pnmto-whatever command. See Section 4.3.1 for information on the netpbm pakage, which allows you to convert .pnm files to all sorts of other types.
![% latex2html id marker 1761
\fbox{\parbox[c]{.8\textwidth}{
NOTE: (See
Section
\ref{scanpdf}
if your goal is a pdf file).}}](img24.png){kind=small}
mkdir scannedpages
cd scannedpages
Note that the ``-y 279'' refers to the length of the page(s) being scanned. 297 conresponds to 11.5 inches. ``-mode 'Lineart''' is good choice for black and white documents. There a lot of options that you can experiment with, all of which are explained in the man page for scanadf.
Once you have located the flatbed scanner (look in the attic) and politely made sure that no one else is using it, you can get started. The scanner makes very little noise when not in use, and has no on/off switch. If logit is on, then the scanner is probably on as well - especially if it is plugged in.
Get a terminal window on logit (either by logging in or ssh'ing from another machine) and type
@:
xsane
For tediously detailed instructions, see 5.3.1
![% latex2html id marker 1809
\fbox{\parbox[c]{.8\textwidth}{
NOTE: you do not ne...
...er. You can operate the scanner by remotely logging in (See
Section \ref{ssh})}}](img26.png)
xsane
The result should be the main xsane dialog window and the preview window shown in shown in figure 5.1
(on the Preview
Window) to see the image of
what's on the scanner. The clarity of the preview image
is not affected by the setting in the main window.
it is just there for cropping and fiddling.
to create a dotted
line box around the part you want. The display in the main window
showing the size of the image in KB and in cm will change as you
adjust the size of the image to acquire.
.
You can determine the format of the image file via the
pull-down menu. The default type .pnm is easily converted
to other types - but if you know what you want, and it's in the
menu well...
@:
xv filename
@: gimp filename
The process by which graphic images of written text are transformed into editable text files is called ``optical character recognition'' or OCR. We do not have an OCR program running locally, but don't despair. There is a site on the web that will do this for you and Gene Hammel says it works pretty well.
Here is Gene's description:
Subject: Re: ocr
http://docmorph.nlm.nih.gov/docmorph/
Contains a good deal of information. Basically, a user must register and have a password, then can log in and do all kinds of graphic and text file conversions (except from pdf to anything else). You upload the file to be transformed, you are informed when it is ready to download, and that's it. It all takes place on the WEB page, in interaction with the file browsing abilities of your own OS.
I had scanned 18pp of typed text at 600 dpi. These were *.pnm files. Each was about 1 MB. I read the instructions on the docmorph page, then sent and received each file separately.
First you click on a link to upload a file. You can type in the path or browse for it, then click submit. After it uploads you get a message saying to wait; it takes 12 seconds to process a page. Then a link appears that allows you to download the file, which appears in your WEB browser. I then saved this page as *.txt; the file naming is automatic. Then I clicked the other link, to upload a new file, and went through the process again.
There were no difficulties whatever. The OCR seems to be perfect although I have yet to read through it. The txt files are about 3 KB each. The only problem is tables; these do not preserve their original format, and I need to inquire of their tech support how to do this.
It might have been simpler if I had catenated all the pnm files and uploaded just one file for conversion. But I do not know what happens to a set of pnm files when catenated, and it might be tricky to get them in the right order if one used a regexp in the cat command line so as to avoid a lot of tedious typing. Something like cat file[0001..0055].pnmallfiles.pnm might work.
In short, I do recommend it.
This used to be complicated, but now it is simple: use FreeNX. To do so you will need to install a free client application on your machine and do a little configuring. Once you have it set up, you will be able to connect easily and efficiently and do just about everything from afar that you can do from 2232 Piedmont.
FreeNX is explained in http://lab.demog.berkeley.edu/LabWiki.
The Demography Lab is accessible only via encrypted connection. Telnet, and rlogin are not encrypted so you cannot use those. If your personal computer runs linux or mac OSX, ssh is probably installed and usable, just open (``Terminal'' on a mac) and shell type
@:
ssh demog.berkeley.edu
NOTE that this is also the command you would use to connect to a server from within the local network - except instead of demog.berkeley.edu you would type the name of the server e.g. tapinos, or coale.
If you are running Windows, below are two other options.
![\fbox{\parbox[c]{.8\textwidth}{NOTE: firewall software often inteferes with ssh ...
...bolical
ways. If things don't work as advertized, try disabling your firewall}}](img33.png)
Several options exist for running ssh from windows. All of them require some installation. Most people (who need a secure shell under windows) use PuTTY. http://www.chiark.greenend.org.uk/~sgtatham/putty/.
![% latex2html id marker 2102
\fbox{\parbox[c]{.8\textwidth}{BUT WAIT: Campus sec...
...e no clue about
what a mac address is or how to find it see \ref{macaddr}.
}}](img34.png)
AirBears. AirBears is run by IST, it is available from all over campus to anyone with a CalNet ID.
DropBox is an application that synchronizes files and directories between machines. You can configure it to do so between your portable machine and the Demography servers. Just find [Applications][Internet][DropBox].
TWO IMPORTANT CONSIDERATIONS If you decide to setup DropBox:
If you have secure shell installed on your home computer, you can use sftp. The sftp program functions somewhat like ftp - which we no longer use as it is not encrypted.
If your home machine runs Linux, you would type:
@:
sftp userid@demog.berkeley.edu
after logging in, you can move around and find stuff using cd and ls. To move a file to the machine you are sitting in front of, type get filename to move a file to the Demography Lab network, type put filename.
Under MSWindows, the ssh which you can download for free (as long as you affiliated with UCB) see 6.2, has sftp application with a reportedly ``easy to use'' interface.
NOTE: sftp is picky about what it sees on STDOUT. If your .bashrc echo's a message, it is possible that sftp will hang after taking your password.
One option for transferring/synchronizing entire directories is rsync. Rsync is free and available on many platforms. It is much more powerful than it is simple to use so be careful. Treat it as you would a circular saw...with all the safety guards removed. Read the man page carefully before experimenting. Your best friend, where rsync is concerned, is the --dry-run argument. Also, note that in order to use rsync from outside of the department, you will need to tell rsync to operate over an encrypted connections. This is done via the -e ssh argument.
Here's a simple example of how to move all the files from a directory called dissertation in your Demography Lab home directory, to directory called backup/dissertation on your home computer.
Assuming your home computer runs Linux, of course:
@:
rsync -uva -e ssh userid@demog.berkeley.edu:dissertation ./backup
Instructions for setting up your portable computer to print to Demography printers are found in http://lab.demog.berkeley.edu under the [Documentation] link.
[printing from windows machines]
The ``mac address'' also goes by the names ``ethernet address'', ``hardware address'', ``station address'', ``ethernet id'' and ``physical address'' and perhaps there are others. It is a 12 digit hexadecimal number generally shown with a ``:'' separating each pair of digits. Since it's hexadecimal, digits include the letters A-F. For example: ``00:0C:76:00:4A:FA''.
In most cases the mac address will be printed somewhere on you network card. If not, your computer can tell you. Under Linux the command is /sbin/ifconfig. The numbers you want will be labeled ``HWaddress''.
Under Windows, if you can find a ``command prompt'', you can type: ipconfig /all. The answer will be reported as ``physical address''. NOTE that there may be more than one such physical address so make sure you get the one associated with the wireless card. (Thanks to Sarah Staveteig for this)
Here is a helpful site with more instructions on how to find your mac address under various OS's: this http://www-dcn.fnal.gov/DCG-Docs/mac.
Be especially vigilant if you have two or more network devices-as most portable computers do. Each such device has its own unique mac address. Make sure the mac address you give us is associated with the device that you plan to connect to the LAN with.
A good place to start is with the Not so short guide available at http://www.demog.berkeley.edu/Refs/lshort.pdf. There are also several pretty good books on using LaTeX. Unfortunately the tend to be rather expensive. There should be a few copies of LaTeX: A Document Preparation System by Leslie Lamport lying about the lab.
Emacs is great way edit LaTeX documents, if you like emacs. By visiting a buffer/file with a .tex suffix, emacs should put you in ``LaTeX'' mode all sorts of good things will follow therefrom. Documentation on Auctex can be found at http://www.demog.berkeley.edu/Refs/auc-tex.ps
For those not inclined toward emacs, there are other choices such as LyX http://www.lyx.org/ or on macs, TeXshop http://pages.uoregon.edu/koch/texshop/
@:
pdflatex filename(.tex)
.
But embedded graphics are lost, so if your document is just text this is quick and easy. But...
If your document has embedded graphics then the two step process below works very well:
@:
latex filename.tex
@: dvips -Pcms filename.dvi
@: ps2pdf13 filename.ps
It is also very effective in almost all cases to convert the .dvi file to pdf using:
@:
dvipdf filename.dvi
.
The command
{\use{hyperref} is a nice package to include in documents that
are bound for the web.
There are two pretty good choices:
A reasonable approach for making tables in LaTeX is with gnumeric. Gnumeric is nice spreadsheet program which can read .xls files and can save them as .tex files.
It is of course, not as simple as one might wish. Gnumeric uses the longtable package so in order to modify what gnumeric does, you will to come to terms with longtable.
Here is a rough procedure for getting a table into a LaTeX document. If your tables aren't too complicated and you are not too picky, it'll probably work. But if the wheels come off, you'll want to have a look at the documentation - there's lots of LaTeX docs at www/Refs/LaTeX-DOCS.
\caption later.
[save as] and select ``LaTeX 2e.''
\usepackage commands that you'll
need to insert in your ``including'' document. That is in the file
that you want the tables to show up in ultimately.
All of the \usepackage and \newlength commands come
between the \documentclass and the begin{document}
command. That is the go in the ``preamble.''
\begin{table}[htbp]
\end{table}
You'll surely want to add a
\caption and a \label to your table. To do so, you need
to edit the file that gnumeric wrote.
Look for the command that looks like:
\begin{longtable}[c]{%
b{\gnumericColA}%
b{\gnumericColB}%
b{\gnumericColC}%
b{\gnumericColD}%
}
and make it look like this:
\begin{longtable}[c]{%
b{\gnumericColA}%
b{\gnumericColB}%
b{\gnumericColC}%
b{\gnumericColD}%
}\caption{Fascinating Table Showing Dramatic Result}\\
\label{tab:fascinating}
NOTE the \\ at the end of the \caption.
\input
directive will act as though the input'ed file is part of the
including document. NOTE that since the input'ed file contains a
table-like float, the table counter will be updated, and the
listoftables directive will recognize it as a table. Also,
since it's a float, the table will
appear where LaTeX thinks it should - which might not be right where
the \inpt directive resides.
The most convenient way to access the stuff on your floppy or CDROM, is to make the contents of the entire disk part of the filesystem8.1 The process of joining a device to the filesystem is known as mounting and can be accomplished manually via the mount command. But generally, the workstation will mount it automatically or nearly automatically and put something on your desktop for you to click on.
![% latex2html id marker 2933
\fbox{\parbox[c]{.8\textwidth}{
What you cannot rel...
...s
\textbf{un-mounting}. See Section \ref{umount} for the why and how of this.
}}](img35.png)
So, you stick your CDROM or floppy or USB device into the appropriate slot and then look for evidence of your new device.
On a good day, your workstation will automatically mount it and there by join it to the filesystem at the approrpriate mount point8.2. It will also probably pop up a nautilus (file manager) window in which you can putz around and click and drag and whatnot.
Well if you've been bad, and you device does not appear as promised you can try the following things:
![\includegraphics[scale=.025]{diskmount}](img36.png){kind=small}
if you
on it you should see a menu
that will allow you to ``mount'' or ``unmount'' the device.
[Prefences][Hardware][Removeable
Drives and Media] for further options.
Though mounting a device is simple and automatic, it is not the only way to access data on floppy disks. See Section 8.6 for an alternative.
The easiest way to un-mount a device is to click
on the
icon (on your taskbar) then select
[unmount] or [eject].
If your are working with the device through the filemanager (nautilus), then should be an unmount or ``eject'' button within nautilus as well.
![\fbox{\parbox[c]{.8\textwidth}{NOTE that unmounting will fail if you are accessi...
...cations that might be accessing the media
\textbf{before} you can unmount it.}}](img37.png)
In the simplest case where you have a standard 1.44MB floppy and you want to put a basic DOS filesystem on it (so you can read the disk on a windows machine) just put the disk in the drive and type:
@:
formatfd
This command is locally aliased to run:
/bin/fdformat /dev/fd0H1440; mformat a:
read on to find out
what this means.
If you only put tar files on floppy
diskettes then the ``high level'' format described below is
irrelevant. It neither helps nor hurts to create it. You
must, however, be sure that your diskette has a low-level
format. Instructions for creating a low-level format are given
below. But if you wanna save time... just run
@: formatfd
like everyone else. Then
@:
tar -cvf /dev/fd0 ./path/to/files
Floppy disks have 2 levels of formatting: a ``low level'' format which determines how the little ones and zeroes are stored on the diskette. This has to do with tracks and sectors and stuff like that. Once the low level format is done, diskette controllers handle it and it becomes very uninteresting to end users. The only thing you need to know is that if you have a raw diskette, you have to make sure this low level format is created. In almost all cases it will be pretty clear if the low level format is faulty, because just about nothing will work.
The second ``high level'' format refers to the filesystem. The filesystem takes care of file names, permissions, directory structures and that sort of thing. For most purposes, you will want your floppy disk to have a DOS or ``FAT'' filesystem on it. These can be read both by Windows and Linux machines. If you only use Linux machines, you may wish to format your diskettes as ext2 but you will probably be the only person on the planet who does this.
If you wish to put an ext2 (linux) filesystem on your floppy disk, use the command:
@:
/sbin/mke2fs /dev/fd0
@: mformat a:
Note that mformat is part of a suite of tools known collectively as mtools. See Section 8.6.
mtools contain a version of just about all of
the DOS commands that can be used to fiddle with floppies just stick
an 'm' in front of the DOS command as in mcopy,
mdir, mcd, mdel and mdeltree. All of these exist and work as you
might not expect with a: instead of /dev/fd0.
Consult the man page (@: man mtools) for a complete list.
Writing CDs is a 2 step process (or a 3 step process if you include going to the store (or Rm 101B) to buy a blank CD-(W)R disk). First a disk image must be created. This is file that contains the image of what will eventually be on the CD. Since you can only write once to CDs you cannot simply add files one at a time. The entire filesystem including the bookkeeping parts that you don't generally realize are there must be assembled before the CD can be written. A filesystem image assembled in this way is usually stored in a file with a ``.iso'' suffix. The second step is to transfer this filesystem image to the CD.
You need only be vaguely aware of these two steps if you choose to make your CDs with nautilus file manager application - aka ``the easy way''.
![\includegraphics[scale=.025]{homedir}](img42.png){kind=small}
or by selecting:
[Applications][System Tools][file Browser]
button.
seventeen or eighteen times.
![\fbox{\parbox[c]{.8\textwidth}{\textbf{\textit{Gotcha:}}If you are using previou...
... to mount it for you. You cannot write to a disk
that is mounted for reading.}}](img41.png)
@:
cd-image.iso is the filesystem image that you want to make. The .iso suffix is optional, but cheap so why not do it.
![\fbox{\parbox[c]{.8\textwidth}{The $<$\texttt{iso-level 3}$>$ arguement tells \...
...options violate
the ISO9660 standard but \lq\lq happen to work on many systems''. }}](img44.png)
cd-image.
Once your disk image is made, you can use wodim to transfer
it to CD - Or you can just
on the .iso file in a nautilus
window and select Write to Disk.
The macho way is to use the wodim command. To transfer a cd
image called cd-image.iso that is in the current working
directory - probably /other/72hours. Don't forget to put your
blank CD/DVD disk in the drive before running wodim.
@:
wodim -v -data cd-image.iso
Once the transfer operation is complete, you can inspect the CD's content by mounting it. Just eject the CD and then push it back in.
NOTE: if you are re-using re-writeable media, you must ``blank'' or erase the disk before attempting to write to it. See 8.7.2 for instructions.
![\fbox{\parbox[c]{.8\textwidth}{If the above command fails, it is probably becaus...
...ttt{wodim dev=/dev/scd0}}
\smallskip
\par
will then solve the problem.
\par
}}](img46.png)
To blank a disk:
@:
wodim blank=fast
On a good day, the drive should now close by itself and the blanking process should start.8.6
On a not so good day, wodim will gag on the above command. To make wodim try harder you can add the -force flag. Ultimately, the following command will do the trick, but it can take quite a while to complete:
@: wodim -force blank=all
Your best (long run) option for word processing is LaTeX - See www.demog.berkeley.edu/Ownersmanual for and impassioned diatribe regarding the virtues of logically oriented text processing.
If you lack sufficient idealism to pursue LaTeX - or if you just prefer something will lower startup costs - the OpenOffice suite is a good options.
OpenOffice (OO) is an open-source alternative to the Microsoft Office software. OO is supported by Sun which directly supports StarOffice - we generally use StarOffice here (since the campus buys it) but I'll refer to it as OO anywya. OO works just about the same way as MS with a few exceptions that might annoy you at first - if you are accustomed to the MS way and are too rigid to consider alternatives.
OO and MS office software are sufficiently similar that one can easily go back and forth between them. OO can both read and write all of MS's secret proprietary formats and even has its own version of the infernal talking paper clip.
It is also possible to run MS Office applications under Linux. We do this with cxoffice See 10. Since MS applications run on Linux only under duress, you can expect some quirks and glitches. Of course, you expect that with any MS software.
It is less frustrating to work with the OpenOffice applications under Linux. OpenOffice is also available for Windows and Mac for free.
If one moves back and forth between OO and MS, one must be a bit careful about equations. OO can display equations created in MS but it is best not to try to edit them. MS does not understand OO's equations at all.
OpenOffice also has a spreadsheet that compares well to MS Excel. As with the word processors there are slight differences, but OO can do everything that MS can and once you get used to the small differences it is not difficult to move back and forth between them.
The first time you run open office, a screen will popup offering to allow you to ``install'' OpenOffice. All the default suggestions are fine so if you're brave you can stop reading here and click away. More details for the more timid are below.
What happens first will depend on whether you have used previous versions of OpenOffice before. If you have you may be prompted to update and old installation - resist this temptation and ``install'' in a new directory.
Eventually the following steps happen:
to go on. (you can decide for
yourself).
. Some lights flash and then
you hit the
and you're done... with the installation
After the setup program exits, you can launch any of the various OpenOffice applications as described in 9.0.6
There are several ways of launching OpenOffice applications.
@:
soffice filename
where filename includes the typical file suffix such as .doc or .xls. soffice will choose the appropriate application to open you document with.
If you wish to start a new document, then you can save keystrokes by launching the particular application that you want to use. The application command names are shown in Table 9.1
|
[Office]. They each have their own nifty
post-literate symbol on the panel.
to select which
application you want.
Crossover Office (or cxoffice) is commercial grade WINE implementation. Its purpose is to allow windows addicts to run their favorite applications under Linux.
It works amazingly well - considering the challenge. In other words, it works, but there are a few bugs.
From the [RedHat][X Windows Applications] you can launch
Word, Excel, or Powerpoint. In nearly all respects these program
behave, under Crossover Office, exactly as the do under Windows. With
just a few additional idiosyncrasies and bugs thrown in.
Currently the following windows applications can be run from your Linux desktop:
More will be added as we gain experience with this system.
Be extremely careful with this command - if you screwup bad things could happen.
@:
cd; /bin/rm -r .cxoffice
@:
/other/cxoffice/bin/cxreset
When it asks permission to proceed, say yes.
echo $PRINTER
@: echo $LPDEST
If the name of a printer is returned from either of these commands, edit your .tcshrc file and delete them. See the Owner's Manual if the phrase ``edit your .tcshrc file'' intimidates you.
If your .tcshrc file did set PRINTER or LPDEST, then you will need to logoff and log back in before you continue.
If the above procedure worked, then printing to the ``wine-postscript-driver'' will launch a printing application which will give you lots of choices as to where to send your print job and how to print it. The first time you see this kprinter application you must change the ``printer driver'' setting from LPD to CUPS. This only needs to be done once (unless you do that radical reset procedure in 10.3.1.
Hereafter, always select the ``Wine-Postscript-driver''. And never try to print MS docs to status. (See 10.3.2).
ps2ps filename - | lpr -Pstatus
The equation editor under cxoffice is prone to crashing but the crashes are not random.
CensusCD 2000 works properly, but the 1980 and 1990 applications
generate lots of errors when you try to create a flat ascii or dbf
file of the data. Mapping works fine - it is only the data exporting
function that fails. In most cases, it writes the file before failing
for what that's worth. But you will need to use
[RedHat][CrossOverOffice][Reset Crossover Office]
to recover control of your desktop after each failure.
Senate Bill 1386, Assembly Bill 700 and common sense dictate that care must be taken with personal financial data stored on computers - especially computers connected to the Internet.
By directive of the Chair, there shall be no systematic collection or storage of protected Information on Demography Department Computers. By Protected Information we mean ``protected information'' as it is defined in the above noted Senate and Assembly Bills. This includes an individual's first and last name in combination with any of the following:
Staff, faculty and students are therefore instructed to remove any and all such information from any database system or file on any networked device in the Demography Department.
By Dec 31, 2005 (or when your account is set up) you will be asked to sign a statement promising to comply with this directive.
The chair also advises that when called upon by outside agencies to provide social security numbers for Demographers, polite resistance is a good strategy. It is often the case that even though the important looking form asks for a social security number, none is actually required.
Sensitive data such as grades and exams which need to be accessed on computers should either be stored on removable media only or be encrypted using a public-key encryption system such as GnuPG with an adequate keysize.
If you choose to encrypt files there are choices to be made about how that should be done.
The program gpg (GnuPG) is the best option for encrypting files. For details on how to use it's many features check the man page:
@:
man gpg
Briefly, gpg has two modes of encryption:
This is not such a big deal if you have also stored that data on a CDrom, under your bed.
If you need encrypt with multiple keys, contact carlm and we'll figure something out. If you just want to do the easy thing see 11.3
@:
gpg -c filename
you will be prompted twice for a pass-phrase - it can contain spaces and it can be long.
You will probably see a message warning you about ``insecure memory''. Don't worry about it unless you like to worry.
When complete you will have a file called filename.gpg as well as your original unencrypted file.
Obviously it would be rather boneheaded to leave the unencrypted file in place once you are sure that the encryption worked. right?
To decrypt the file you just encrypted type:
@:
gpg -d filename.gpg > filename.decrypted
You don't need to call the resulting file filename.decrypted any filename will do. But existing files by the same name will be overwritten.
Before deleting the unencrypted file you may wish to check that this process really worked. Here' how:
@:
cmp filename filename.decrypted
cmp returns nothing if the two files are identical. If they are not - it tells you so.
Now, you can safely erase the unencrypted file - if you can remember the password.
L
The short answer is ``lots'' - but there is a more complicated answer. The demography system maintains 3 distinct types of disk space for distinct types of uses:
These intermediate results / temporary files belong in /72hours where they are not backed up an are deleted after 72 hours of disuse.
If you are using large data files, and you politely store them in a compressed state (See 12.4 for advice on this) temporary directories like /72hours are generally good places to uncompress.
NOTE: /72hours are not networked. That is, /72hours on coale is not the same as /72hours on tapinos or census or whatever. They are all different, and to move stuff between them you need to use a program like sftp.
|
The quota system is designed to keep users from ``carelessly'' consuming huge amounts of disk space. Of course this is far more complicated than simply refusing to do anything for users when their allotment of space is used up. That would be mean. Instead when a user exceeds her soft limit for a particular filesystem, the quota system begins a count down that lasts a period of days. While that countdown is in progress, the user can operate as if nothing were wrong - but she will get warnings when she either logs onto a server or reads her email.
When the countdown (or ``grace period'') is complete, the system then imposes the only penalty that gets the users attention: no more disk writes until disk use falls below the soft limit. Read access is still permitted, and so is file removal.
There is also, a hard limit over which a user's disk use can never go. This is designed to stop terrorist processes from filling up entire disks.
We set the hard limits, soft limits and grace periods independently for each filesystem. Home directories, /data/commons, /Sastemp and /other/72hours all live on different filesystems and therefore have different soft limits hard limits and grace periods.
If you are the sort of person who prefers to shape her own destiny, then it behooves you to periodically monitor the growth of that garbage heap which is your home directory. Since your stored email as well as your email inbox count against your Home directory quota, things can get quickly out of hand.
@:
quota -v
shows your disk use on all relevant filesystems. If you are over your soft limit, it tells you how much time remains before you will be punished.
A good place to start your search for trash is:
@:
findtrash
which is an alias that executes: du -sh .[a-zA-Z]* * |egrep -e '[0-9]M' | sort -n. which (obviously) produces a listing of directories ( including ``.dot'' directories) sorted by space consumed. Note this alias takes a minute to run
For more precise tasks, the command du (for ``disk use'') will produce a directory by directory listing of amount of space used in each of directory. With a -k flag it will give you the data in Kilobytes, on linux machines the -h flag gives you file and directory sizes in ``human'' readable units.
@:
du -k |less
Will give you the output page by page.
![\fbox{\parbox[c]{.8\textwidth}{NOTE: files with names like \texttt{core} or
\tex...
...you can safely delete them - and you should
as they tend to be \textbf{huge}.}}](img50.png)
Once you locate a directory with lots of stuff in it, you can use ls and sort to find the big files. The command
@:
ls -s | sort -n
prints each file precede by a number indicating it's size, the sort -n order the output numerically.
Compressing files can save you a lot of space. Data files in particular can sometimes compress by as much as 90%. There are several different programs/algorithms for compressing files. Three common ones are:
bzip2 is the fastest and most powerful, but it is not yet standard so if you need to uncompress your data later on an Amiga, bzip2 might be a bad choice. GNU zip, gzip offers a good compromise between universality and power. compress is not much used anymore.
To gnu compress a file type the following:
@:
gzip filename
The result will be a file called filename.gz. The original file will not be harmed - you must remember to remove filename yourself, otherwise file compression will just have resulted in additional disk usage.
There are two ways to uncompress a file - one deletes the compressed file and one does not.
To uncompress leaving the compressed file in tact, use zcat:
@:
zcat filename.gz > newfile
This will create a file called newfile leaving filename.gz in tact. This is a useful thing to do if your datafile is so large that it takes a long time to compress and you are uncompressing onto a temporary directory (See 12.1 for details).
To uncompress and remove the compressed file at the same time type:
@:
gunzip filename.gz
A good way to compress an entire directory and all it's subdirectories is with the tar command. This creates a single ``tape archive'' file which can later be ``un-tarred'' to reproduce the original directory structure. The tar command has lots and lots of options - so check out the man page ok?
Here is a simple example:
@:
tar -czvf dissertation.tgz ./dissertation
Here ./dissertation is a directory holding lots of files and perhaps many subdirectories as well. The original directory is not changed by this action. So if you goal is to save space, you will need to erase the uncompressed original. dissertation.tgz is a compressed tar file which can be moved, mailed, renamed, or erased just like any other file. But unlike any other file, it can also be untarred to reproduce the original directory - perhaps in a different place, on a different computer. Here is the command to untar:
@:
tar -zxvf dissertation.tgz
The file dissertation.tgz is not destroyed or changed in this process. However, untarring can overwrite preexisting files. The command above will recreate ./dissertation exactly as it was when you tarred it. If, therefore, you tar -x ... in the same directory as you tar -c'ed earlier and in between time, you improved but did not rename, some files in dissertation then the untarring will overwrite the new improved files with the old files of the same name. This is all fine of course, As long as you expect it.
The purpose of tape backups is to allow us to recover from nuclear attack, earthquake, typhoon or firestorm. It is vital that users realize that system backups are worthless for reproducing scientific results from a project more than a few days old. Backups are designed to allow us to recreate the filesystem as it (mostly) was on a very recent day. If a disk fails, we need to be able to replace the disk and its contents as of yesterday and that is the goal of system backups.
In order to concentrate resources on backing up the most critical files, we employ several strategies of which a good scientist should be aware:
Besides saving resources, the system we have in place encourages good programming practice, reproducibility of results and thus good science.
Knowing as you do now that large data files are vulnerable to hardware failure and terrorist attack, you have no doubt resolved to follow the practices outlined in Demog 213:
Reproducing results that you generated 2 years ago is a task for which system backups will be completely useless. In order to be able to reproduce your work you must make an archive of your project at the time it is complete. That archive may contain files from all over the filesystem and, if you're clever, it will contain some notes on how you did what you did so you (or your RA or your biographer) can do it again after you are dead.
Archives are the responsibility of the only person who cares about and understands the work you do.
There are several choices regarding archiving. For projects requiring less that 750MB, CDROMs are a good choice. DVD's hold about 5GB and are just as easy to write as CDs. See 8 for instructions on how to use the various removable media.
This document was generated using the LaTeX2HTML translator Version 2008 (1.71)
Copyright © 1993, 1994, 1995, 1996,
Nikos Drakos,
Computer Based Learning Unit, University of Leeds.
Copyright © 1997, 1998, 1999,
Ross Moore,
Mathematics Department, Macquarie University, Sydney.
The command line arguments were:
latex2html -split 0 -local_icons faq
The translation was initiated by Carl Mason on 2012-08-21
To do this go to
[System][Preference][Hardware][Removable
drives and media] and clear the ``mount removable media when
inserted'' check box.
