Start · Persons · Teaching · Research · Conferences · Guests · Contact
Lasse Kliemann

Short HOWTOs for frequently occuring technical problems and tasks

Some information here is only relevant for members of our research group.

Von dieser Anleitung gibt es auszugsweise auch eine deutsche Version.

Corrections and additions are welcome, preferably as diff for the sources of this page. (The sources are written in my own HTML preprocessor language. In case you are interested in learning more about it, please contact me.)

Table of contents

The dates give the date of the last modification.

Some notation

~
A tilde followed directly by a username, e.g., ~lki, stands for the home directory of this user. If followed by nothing or by a slash, it stands for your home directory. You can use this in commands that you enter. For example, the following command will change the current working directory to the directory public_html located in your home directory: cd ~/public_html
MUA
Mail User Agent. E.g. mutt, mailx, elm, pine.
abc
If code is set in this style, it usually has to be replaced appropriately for an application. For instance, I write: svn add file. This means that file has to be replaced with the name of that file on which this command should operate.
abc
If code is set in this style, it usually shows a complete command for the shell. It can be entered by the user, possibly after having replaced certain parts of it appropriately for the application at hand. Please also see the previous item. This style is also used for blocks that are set off from the main text and contain a series of commands or code snippets.

How do I work with a unix-like system in general?

[Last modified on 2007-12-11.]

Please see the following references:

How do I login to another machine?

[Last modified on 2007-10-22.]

Likely, you were given a string of the form username@hostname, or a username username and a hostname hostname separately. Maybe you were also given a password. To log in using SSH do:

ssh username@hostname

Input your password when asked for it. In many cases, a different authentication method was set up, or can be set up, which does not require you to input your password every time. Read more about SSH keypairs if interested.

If you need to run X applications, read how to forward X connections.

How do I copy files from one machine to another?

Limited solution: scp

Copy the local file file to remote host host into the home directory of user username (do not forget the colon at the end of the command):

scp file username@host:

...or into a subdirectory dir located in the home directory of user username on the remote host:

scp file username@host:dir

Copy the file file from the remote host to the local directory:

scp username@host:file .

The most important options are:

-p
Preserve certain file attributes, like modification time and permissions (usually a good thing).
-r
Copy directory trees.

So, the following copies the local directory tree dir to the remote host host, into the home directory of user username, preserving file attributes:

scp -rp dir username@host:

Powerful solution: rsync

rsync has many advantages over scp. One of them is that if some files already exist on the target system, it only transfers the differences between source and target system, hence speeding up the transfer. It can also be made to skip files that have more recent modifications on the target system. For example, copy the directory tree rooted at dir1 to the remote directory tree rooted at dir2. The following command does the job, while also preserving file attributes, copying symbolic links as symbolic links, and such (this is due to the switch -a):

rsync -a dir1/ username@host:dir2

Note that the trailing slash behind dir1 is mandatory. Without the slash, the target directory would be dir2/dir1 on the remote host.

Files present in dir2 and not present in dir1 are not touched in this case. If you would like such files to be deleted, add --del to the command.

Files present in dir2 and present in dir1 will be overwritten in dir2 with those in dir1, unless you specify the -u switch. With this switch in place, files in dir2 will not be overwritten provided their last modifications are more recent those of the corresponding files in dir1. This has some special issues in case destination and source files of the same name have different types (e.g. a regular file vs. a directory); please see the manual page for details befor relying on -u.

rsync can also be used locally, e.g.:

rsync -a --del dir1 dir2

For more information, see the rsync manual page.

How do I work with tar archives?

[Last modified on 2007-11-03.]

Create an archive

The following creates an archive named archive.tar containing the files file_1, file_2, ..., file_n

tar cf archive.tar file_1 file_2 ... file_n

The concerning files may be regular files or directories. In case of a directory, the whole subtree rooted at this directory is archived.

To have a compressed archive, afterwards do

gzip archive.tar

This removes the existing archive and writes the compressed version under archive.tar.gz. If you would like the file to have the ending .tgz, just rename it:

mv archive.tar.gz archive.tgz

If someone explicitely asked for a .zip file, then this is not the right procedure. You must use the zip command instead of tar and gzip then. This is not covered here.

Extract an archive

Extracting an archive is slightly more delicate than creating one, since extraction may overwrite existing files. Unless you know exactly what is in the archive, change to an empty directory before extracting the archive. Now, Solaris tar even under that precaution may overwrite existing files (e.g., if the files in the archive contain .. as path elements). Therefore, use star or gtar on Solaris for extraction. The command syntax is the same for both of them and also the same for the normal tar. I use tar as the command in the instructions below; please replace accordingly when working under Solaris.

If the archive comes compressed, uncompress it first:

gunzip archive.tar.gz

or, in case of a .bz2 archive:

bunzip2 archive.tar.bz2

This yields archive.tar. To see what is in the archive, do

tar tf archive.tar

The archive is extracted with

tar xpf archive.tar

This restores most of the files' metadata as it is stored in the archive, including the time of last modification. This usually is a good thing. If not desired, omit the p, i.e., do

tar xf archive.tar

How do I use subversion?

[Last modified on 2007-11-06.]

Once, in the beginning, obtain a working copy from the repository. This operation is called a checkout. The repository is a database on a server, which contains all files and metadata concerning a specific project. The repository is identified via an URL. The URL is given to you by me.

To check out a working copy from the URL URL, do:

svn co URL

This creates a new directory. If the URL is svn+ssh://account@example.com/dir1/dir2 then the directory dir2 is created by the checkout. The working copy is everything below dir2.

A checkout is only necessary if you do not have a working copy already! People often try checkouts over or inside existing working copies, which has unexpected results. (It is, of course, possible and often useful to have several working copies on different accounts, or on different machines –one at home, one at work, one on the notebook, etc.–, or on the same account in different directories.)

Once you have a working copy, in a nutshell, it's essentially all around the following two operations:

The svn commands have to be issued while you are in the working copy. Use cd to change into the working copy.

In more detail, the workflow goes as follow:

Usually, the final action should be svn ci, i.e., a commit.

If you are to move an already existing set of files under the control of subversion, simply copy the files into the working copy using cp and then use svn add on them.

If files that are under version control have to be copied, moved (including renaming), or removed, do not use cp, mv, or rm directly. Use svn cp, svn mv, and svn rm instead.

For more information, please see the slides of my talk. You can also read the subversion book and the subversion FAQ.

Tips and tricks using subversion

[Last modified on 2007-10-09.]

Close all editor buffers before an update!

Suppose you edit a file in your working copy. The editor maintains an internal copy of the file's contents. If the file is modified on disk (e.g., by an update), the editor's copy is not changed, meaning that it still reflects the file before the modification. Writing the file to disk from the editor hence reverts the modifications.

This is usually not what you want.

Frequently do updates and commits!

An update gives you the chance to adapt your work to the changes made by others. A commit gives others the chance to do the same regarding their work and your changes.

Use svn status and svn diff!

These commands cannot be issued too many times. svn status gives you an overview of the status of your working copy. See svn help status for an explanation of the output.

svn diff shows your changes made since the last commit in detail.

Use svn cp, svn mv, and svn rm instead of cp, mv, or rm directly!

The reason is simple: to allow subversion to keep track of copied, moved, or removed files. This has many advantages, one of them being that (when copying or moving) the history of the file does not get lost, but is copied or moved along with the file itself.

Write short lines in your source code!

When using LaTeX, HTML, or similar, a single line-break does not influence the final presentation of the text. Hence you can use line-breaks to make it easier to edit the text, or generally to work on it, for that matter. One way is to insert a line-break (pressing the enter key) roughly after every sentence or clause.

svn diff gives more readable output with short lines, because files are compared linewise. Short lines also increase the chances that changes can be merged automatically, and they make it easier to resolve conflicts.

By the way, it is also more convenient to edit a file with short lines, please see the note on text editors.

Examine changes made to files in the past!

svn log file shows you the log of file. This includes the information when the file was modified and by whom, and also any log messages recorded.

svn diff -rn:m file shows differences between revisions n and m. Deleted lines are marked with a minus sign in the first column, and new lines are marked with a plus sign. All other lines give the context of the changes.

How do I make subversion stop opening an editor every time I commit?

[Last modified on 2007-10-25.]

For Bash or Korn Shell

Add the following shell function to your ~/.profile (or your ~/.bash_profile if that does not work) and restart your shell. From now on, svn ci will not slap an editor into your face anymore. To still get the normal behavior, do svn commit, i.e., use the long form of the command. 

if which svn > /dev/null 2>&1; then
   svn_command_tmp_=`which svn` &&
   eval 'svn() {
      case "$1" in
         ci) shift && '"$svn_command_tmp_"' commit -m "" ${1+"$@"} ;;
         *) '"$svn_command_tmp_"' ${1+"$@"} ;;
      esac
   }'
   svn_command_tmp_=; unset svn_command_tmp_
else :; fi

For C Shell

Add the following to your ~/.cshrc file: 

alias svn /home/lki/command/svn_nolog

and restart your shell. From now on, svn ci will not slap an editor into your face anymore. To still get the normal behavior, do svn commit, i.e., use the long form of the command.

The above solution, with the path name to my home directory, will work only on the Sun system of our department. However, everyone may download the svn_nolog shell script, put it somewhere on his or her system, and adjust the path in the alias appropriately.

How do I generate an SSH keypair?

Do you have files ~/.ssh/id_rsa and ~/.ssh/id_rsa.pub? If so, maybe you do not need to generate a new keypair, because you have one already.

To generate a new keypair, do:

ssh-keygen -t rsa -b 4096

and answer the questions. To have a key without passphrase (usually the best choice) and to save it to the default location, just press the enter key on all questions.

To send the key to me, do:

mail lki@informatik.uni-kiel.de < ~/.ssh/id_rsa.pub

Or use your favourite e-mail client to send the file as an attachment. You may also copy this file to another account first (using, e.g., scp or rsync), from where you can send e-mails. This file does not contain secrets. You should, however, keep your ~/.ssh/id_rsa file confidential; there is usually no need to copy or send it around, just leave it where it is.

If you would like to use the key to give yourself easier access to an account acct that you already have access to (maybe via a password), add the contents of ~/.ssh/id_rsa.pub to the file ~/.ssh/authorized_keys on the account acct.

Note: Maybe a shorter keysize than 4096 fully suffices. Maybe DSA keys are better. If you know something about it, please contact me.

How do I set up my personal homepage?

[Last modified on 2007-11-02.]

If your username is foo, then a file ~/public_html/file can be accessed by the public under the following URL, provided that certain permissions are set appropriately:

http://www.informatik.uni-kiel.de/~foo/file

If someone just types

http://www.informatik.uni-kiel.de/~foo/

into his or her browser, the file ~/public_html/index.html is served.

Permissions

The permissions concern the home directory and the public_html directory as well as the files in the public_html directory. I suggest the following for the home directory and the public_html directory:

chmod 711 ~ ~/public_html

Be aware that after this, files in your home directory that have some permissions for the group or for others will really be accessible by group members or by others. If you are unsure and want to prevent exposure of secret information, first do:

chmod -R go-rwx ~

and then set ~ and ~/public_html to mode 711 as shown above. To learn more about permissions, type unix file permissions into a search engine and navigate to a search result of your choice.

Now, the files under ~/public_html which you would like to share with the rest of the world have to be made world-readable. To accomplish this for a file ~/public_html/file do:

chmod o+r ~/public_html/file

Creating your index.html

To give all personal pages the same look, I provide you with stylesheets, header, footer, etc. It is easy for you to use them. The following commands will overwrite files Makefile and index.wml in your public_html directory (if they exist).

cp /home/discopt/website/personalhp/dist/Makefile ~/public_html

cp /home/discopt/website/personalhp/dist/index.wml ~/public_html

Now, cd ~/public_html and fill up index.wml with your contents. Leave the stuff at the top and the bottom of the file intact, but fill in values for title, name, and email. Usually, title and name will both be your name, and email will be your email address. When writing the contents of your page, please also consider notes on markup.

The following steps will overwrite an existing style.css, index.html.tmp, and index.html. So, if you are converting from a previous index.html to a new one, make a copy (under a different name) of the previous one first! Then, while in ~/public_html, do:

rm index.html

umask 0077

gmake

The last step generates the new index.html from index.wml and the header and footer provided by me. It also copies new style.css and Makefile to your ~/public_html. Because you set the umask before, the new file index.html will not be readable by others. This is good, because you first want to see if the file is allright before making it available to the public. To see this file being rendered, open it in a webbrowser. If using Firefox, type Ctrl-o and choose the file. If using a text-based browser like w3m, do

w3m index.html

If everything is okay from your point of view, please take some time to ensure that the HTML code really follows the specification. To do so, go to Validome and upload the file index.html via the webform. The validator will point out errors to you. Correct errors by modifying index.wml and then typing gmake again (while in ~/public_html). Then re-validate the file; usually, to this end you only have to reload the last page of the validator, meaning you hit Ctrl-r in Firefox or Shift-r in w3m.

When everything is okay, make the file available to the public:

chmod o+r index.html

Just to make sure, recheck the validity of the code by following the links to the three validators at the bottom of the page. (If using a graphical browser, like Firefox, this means clicking on the images.)

Please repeat gmake; gmake in your ~/public_html at least every month in order that your personal homepage will reflect the latest changes made to the design of our website. Remember to never change index.html manually, since your modification will be overwritten by the next make. Always modify index.wml instead. The same goes for style.css and Makefile itself; they both are overwritten if you run make.

Markup

Photo

To have a photo and contact information side-by-side, you can use the following code in your index.wml: 

<img class="portrait" src="photo.jpg" alt="Photo of ..." //>
<div style="min-height:250px;">
...contact information...
</div>

Replace the 250px with the height of the photo, or more. If you know a better way to do it (not using tables!), please let me know.

Tips and tricks

How can I let my MUA append a signature automatically?

Most MUAs look for a signature in ~/.signature and append them to all you outgoing mail automatically. This has nothing to do with cryptographic signatures.

How do I cryptographically sign and/or encrypt my e-mail?

[Last modified on 2007-11-02.]

This short howto covers some of the most important basics. After reading this, you should be able to deal with encrypted and signed e-mails. Still, a deeper understanding of what is really happening could be helpful. Please do also refer to the GNU Privacy Handbook.

However, you usually do not have to care about the so-called Web of Trust. Beware that many security howtos emphasize its importance. It is not that important; moreover, it brings huge privacy problems. Some important weaknesses of the Web of Trust will later be addressed in this howto.

Public and private keys

You need a keypair, consisting of one private and one public key.

The public key can be used by others to encrypt messages destined for you. The public key can also be used by others to test whether a given signature was made with your private key.

The private key can be used by you to decrypt messages that were encrypted with your public key. The private key can also be used to make signatures.

Create a keypair

A keypair is created with the program gpg. The creation is an interactive process; gpg will ask several questions, you answer them. The following is a walk-through. It will actually create two keypairs, one RSA keypair and one Elgamal keypair. You do not have to worry about what this means.

It all starts with the command

gpg --gen-key

For the type of the key, choose RSA (sign only) and a keysize of at least 3072. How long the key shall be valid is up to you. I usually recommend something between 3 and 5 years.

Real name, email address and comment are also up to you. You do not need to fill in your real name, you can also choose a pseudonym - some name you would like this key to be associated with. The same goes for the email address, and you may also use the comment field for whatever you like.

Next is the passphrase. A passphrase should be at least as good as a password, meaning, at least 8 characters long, not based on a dictionary word, and containing upper-case letters, lower-case letters, and numbers.

That was the first phase of the creation process. In the output you find a line similar to this:

pub 3072R/1A8DB757 2007-10-30 [expires: 2011-10-29]

The identifier after the slash, 1A8DB757 in this case, is your key ID. When you specify the key ID, you should prefix it with 0x, that is, a zero and the letter x. So, the correct way to specify the above key ID would be: 0x1A8DB757. Whenever in the following commands or examples, a key ID occurs, it is recommended that the key ID is prefixed in this way. The next step is:

gpg --edit-key key_ID

At the prompt, type addkey, then input your passphrase (chosen in the previous phase). Next, choose Elgamal (encrypt only) and a keysize of at least 3072. For the expiration, the same comments as before hold. When the key generation is finished, type save at the prompt. This brings you back to the shell, and your keypairs are ready.

Note to crypto experts: Maybe it is better to choose a DSA key as the first one instead of an RSA key. The reason why I choose RSA is that it allows much larger keysizes than DSA. Please let me know in case you have reasons why a DSA key would still be better. (This request is online since November 2007.)

Publishing your public key

The following writes your public key to a file pubkey.asc

gpg -a --export key_ID > pubkey.asc

You can publish it by e-mail, put it on your website, etc.

Collecting public keys of others

If you received or downloaded a file containing a public key, import that key:

gpg --import file

If you only have the key ID of the desired key, you can try to fetch it from a keyserver:

gpg --recv-keys key_ID

Beware that if you wanted a key that belongs to a specific person, at this point you do not know whether the key that you just imported really belongs to that person you have in mind – even if the key bears a name under which this person is known to you. (Belongs to here means that the person controls the corresponding private key.)

How can you make sure the key belongs to the expected person? There are two easy ways:

The fingerprint is a human-readable sequence of characters, short enough that one can exchange it via phone or in printed form. The fingerprint practically identifies the key, meaning that it is impractical for an attacker to create a key with a specific fingerprint. To see the fingerprints of all keys stored on your account, do:

gpg --list-fingerprints

Someone might explain to you that the Web of Trust is a way to achieve certainty that a retrieved key belongs to a specific person. This is not entirely true. The Web of Trust, in its present form, can – at the best – give you certainty that a specific key with name N and e-mail address M belongs to a person who has that name N written in his or her identity papers (which were issued by the authorities) and has control over the e-mail account behind the address M. That does not necessarily answer the question whether the key belongs to that specific person that you expect. One major reason for this is that names in identity papers are not unique identifiers; several persons may have the same name. Using the e-mail address as identifier is better under the aspect of uniqueness, however it requires that you can identify the person that you have in mind by the e-mail address M specified in the given public key.

The personal (or phone based) exchange of public keys or fingerprints in many cases is much easier and much more to the point than trying to use the Web of Trust. Moreover, sometimes you do not have any specific person in mind when you retrieve a public key; you just know that there is someone with whom you would like to communicate. Then there usually is no point in checking any kind of validity of the retrieved public key. The key then is useful to you as it is: you can use it to check signatures of incoming e-mails and hence determine that all these e-mails were written by the same person. And you can use it to encrypt outgoing e-mail to that someone you wish to communicate with.

By the way, here is my public key.

Connecting it to your MUA

The following is a description for the mutt MUA. I recommend using mutt, not only because it operates well with encryption and signatures, but also because it generally is a nice MUA. Put this file somewhere in your home directory, e.g., save it as ~/.mutt/gnupg. In this file, replace (at three locations) the string KEYID with your key ID. Then put the following line in your ~/.mutt/muttrc (or your ~/.muttrc, whichever of the two you use):

source ~/.mutt/gnupg

Restart mutt. If you receive an encrypted e-mail, mutt will ask for your passphrase and then show the decrypted contents. The passphrase will be remembered by mutt for one hour, in order that you do not have to type it again and again for each and every encrypted e-mail. To make this intervall shorter or longer, change the value of pgp_timeout in ~/.mutt/gnupg.

Upon sending an e-mail, mutt gives you a choice of whether to sign the e-mail, encrypt the e-mail, or both. Making a signature does not require any further configuration and is almost never wrong (unless you intend to later deny authorship for this e-mail).

Encrypting the e-mail, of course, requires that you have the public keys of all its recipients. If you have, you can choose to encrypt the e-mail. mutt may be able to determine which public key to use to encrypt the e-mail for each recipient. If not, it will ask you. To prevent that question in the future, tell mutt explicitly to associate that e-mail address with that public key by inserting a line of the following form in your ~/.mutt/muttrc (or your ~/.mutt/gnupg, if you like):

pgp-hook recipient_email_address recipient_key_ID

What text editor can I use?

[Last modified on 2009-12-08.]

I recommend using VI-Improved (vim) or Emacs (emacs). VI-Improved has great support for line-oriented operations.

When using LaTeX, HTML, or similar, a single line-break does not influence the final presentation of the text. Hence you can use line-breaks to make it easier to edit the text. One way is to insert a line-break (pressing the enter key) roughly after every sentence or clause. You can then fully benefit from editors with a good support for line-oriented operations, because most editing tasks will be the re-arranging or deletion of whole lines. Give it a try!

There is a quick reference for the VI-Improved. If you have trouble printing out the PDF version, try this version (gzip compressed) instead.

You might also want to take a look at Seven habits of effective text editing or this more recent presentation by Bram Moolenaar.

I connected to the system via SSH, and now many programs won't work!

[Last modified on 2007-05-20.]

Likely, the reason for this is that these programs require X access. If your terminal has an X display, you can try to route the X application to your terminal by giving the -X switch to ssh.

ssh -X username@hostname

Beware, however, that this might be very slow, depending on your connection. If you just need to edit text files, a text-based editor might be a better choice. The VI-Improved is text-based if you start it in the usual way

vim file

The Emacs can also be started text-based, in its so-called no window mode

emacs -nw file

How do I access scientific journals from home?

[Last modified on 2009-12-07.]

The university maintains contracts with several publishers in order that journals can be accessed from the university network. When you are at home, you do not have this access since you contact the publisher's webserver from outside the university network. If you have a shell account at university, you can however use SSH as a proxy. On your local computer, i.e., your computer at home, do:

ssh -D localhost:8080 username@hostname

This is how you usually access your shell account, but with the additional option -D localhost:8080. Then direct your webbrowser to use your local computer as a SOCKS proxy. In Firefox, this is Edit, Preferences, Advanced, Network, Settings. Then fill in localhost as SOCKS Host and 8080 as Port. If you then access a website from Firefox, it appears as if the request originated from the university computer where you logged in with your shell account.

Caution: any process on your local computer can use the proxy. If you share your local computer with others (who might be logged in over the network), this can become a security problem.

You should terminate the SSH session as soon as you do not need the proxy anymore. You then have to switch Firefox back to No proxy mode, available in the same configuration dialog as the proxy settings.

How do I execute some program that I downloaded or that someone sent me?

You should only execute programs from trustworthy sources. Executing a program means giving complete control over all your files on the current account to that program.

If you are sure you would like to execute the program, there are two ways:

For all this to work, the program file must be executable. If it is not, change the permissions accordingly

chmod u+x prog

The PATH environment variable

See its current value:

echo $PATH

It is a colon-separated list of directories. Upon you typing in a command prog, the shell looks for an executable file called prog in those directories in the given order. The first one found is executed. (In fact, shell functions, built-ins, aliases, etc. are also relevant to this process, but for this explanation this may be neglected.) Check whether in your PATH there already exists a directory which is located under your home directory. If so, it probably is the best idea to just put the program file into that directory. If the program file is called prog then just typing prog will execute it after you put it there. (If using the Z Shell, type rehash once to make this work, first.)

If the PATH does not yet contain a directory under your home directory, you have to augment the PATH variable by some such directory. Assume, you choose ~/command for this. If using the Bash or the Korn Shell, put the following line at the end of your ~/.profile

PATH="$PATH":"${HOME?}"/command

If using the C Shell, put the following line at the end of your ~/.login

setenv PATH "$PATH":"${HOME?}"/command

The new settings take effect when you login the next time.

How do I update my TeX Live installation?

[Last modified on 2011-07-11.]

Once, you have to set a repository: 

tlmgr option repository http://mirror.ctan.org/systems/texlive/tlnet

From now on, an update is performed by the following two steps: 

tlmgr update –self
tlmgr update –all

How do I include an illustration in a LaTeX document?

[Last modified on 2009-12-06.]

First choice is to use TikZ. The manual, including tutorials, is in a file pgfmanual.pdf on any recent TeX Live installation. It is also available online.

If you prefer Xfig instead, the rest of this section is for you.

Simple case: no text to be interpreted by LaTeX

If the illustration does not inlcude any text that must be interpreted by LaTeX, it should suffice to export it to EPS, say name.eps, and then include this file with the \includegraphics command of LaTeX. For example:

\includegraphics{name}

Do not give the ending .eps here. LaTeX will automatically pick up the EPS file when producing Postscript output. If producing PDF output with pdflatex, the EPS files will automatically be converted to PDF files if you also specify the -shell-escape switch to pdflatex.

General case

If the illustration contains text to be interpreted by LaTeX, as it is the case with math formulas, a combined format is helpful. All text that shall be interpreted by LaTeX must be marked as special in Xfig.

I provide a script fig2combined to assist you in creating the combined format. Just call

fig2combined name.fig

This will create three files:

Include the LaTeX part in your document. To scale the illustration, use \resizebox, e.g.:

\resizebox{350pt}{!}{\input{name.tex}}

This will resize the picture to a width of 350pt and adjust the height keeping the aspect ratio.

If modifications to the text inside of the illustration have to be made, you do not need to do it in Xfig. You can modify the LaTeX part name.tex instead; fig2combined will not overwrite it upon successive calls. To have this file recreated by fig2combined, simply delete it and then call fig2combined.

If you change the positions of the text elements, it will be necessary to have name.tex recreated (in order that the new coordinates are included), and this will destroy your hand-made changes to it. An alternative is not to change the positions of the text elements via Xfig, but to modify their coordinates in the name.tex by hand. This sometimes is preferable anyway, because it gives you more precise control.

However, in case frequent changes have to be made using Xfig, it can become a nuisance having to modify name.tex (which usually means modifying or inserting text) manually every time after it has been recreated by fig2combined. The solution is to introduce an additional level of indirection. In Xfig, place to-be-considered unique symbols (e.g., _text_a, _text_b_, ...) in places where later some text shall appear. These symbols will appear as-is in the name.tex file. You can use some automated means to have them replaced by the actual text you want, e.g., you could use a sed script.

How do I print slides created with the LaTeX Beamer class?

[Last modified on 2007-11-27.]

Most likely, you printed your slides and they appear very tiny on the paper. There are several ways to remedy this.

Use the pgfpages package

Put the following in your LaTeX code: 

\usepackage{pgfpages}
\pgfpagesuselayout{2 on 1}[a4paper,border shrink=2mm]

This puts two slides on one sheet of paper, using up all available space. If you would like to only have one slide per sheet (does anyone know an easy way to also rotate the slides in order to use the paper in landscape mode?), instead do: 

\usepackage{pgfpages}
\pgfpagesuselayout{resize to}[a4paper,border shrink=2mm]

pgfpages is available on any recent TeX Live installation.

Use pdftops

Create a PDF version of your slides, say slides.pdf. Then do

pdftops -expand slides.pdf slides.ps

and print the resulting slides.ps. Note that pdftops is not the same as pdf2ps.

pdftops is available on the Suns of our department.

How do I learn to program in C or C++?

[Last modified on 2007-12-11.]

Please see the following references:

More in-depth reading:

How do I write good LaTeX code?

[Last modified on 2009-12-08.]

See also the LaTeX hints by Barbara Langfeld and another comprehensive talk by me. 

\begin{equation*}
n \leq m^2 \enspace.
\end{equation*}

Here are some useful packages. I try to keep the links to the documentation up-to-date, but if in doubt, please consult the documentation on a recent TeX Live installation. 

\let\newfloat\undefined

\usepackage[compatibility=false]{caption}
This is not recommended by its author, but worked well for me so far.

How do I get those umlauts working in LaTeX?

[2009-12-06]

Let us use an encoding that is understood even by older systems, i.e., ISO-8859-1. We tell LaTeX to use this encoding: 

\usepackage[latin1]{inputenc}

The encoding can also be changed after the inputenc package was loaded: 

\inputencoding{latin1}

Your editor might choose a different encoding per default, often this is UTF-8. If you suspect that this is the case, then save your file, then convert it to ISO-8859-1: 

iconv -f utf-8 -t iso-8859-1 file.tex > tmp.tex
mv tmp.tex file.tex

Upon loading the file again, the editor should recognize the new encoding and use it henceforth.

last modified on Thu May 10 20:53:02 2012, Lasse Kliemann
Valid XHTML 1.0   Valid XHTML 1.0 Strict   Valid CSS!