this version only uses shell builtins

tail() { thbin="/usr/bin/tail"; if [ "${1:0:1}" != "-" ]; then fc=$(($#==0?1:$#)); lpf="$((($LINES - 3 - 2 * $fc) / $fc))"; lpf="$(($lpf<1?2:$lpf))"; [ $fc -eq 1 ] && $thbin -n $lpf "$@" | /usr/bin/fold -w $COLUMNS | $thbin -n $lpf || $thbin -n $lpf "$@"; else $thbin "$@"; fi; unset lpf fc thbin; } This is a function that implements an improved version of tail. It tries to limit the number of lines so that the screen is filled completely. It works with pipes, single and multiple files. If you add different options to tail, they will overwrite the settings from the function. It doesn't work very well when too many files (with wrapped lines) are specified. Its optimised for my three-line prompt. It also works for head. Just s/tail/head/g Don't set 'thbin="tail"', this might lead to a forkbomb.

I've been looking for this for a long time. Does anybody know how to do this in dash (POSIX shell)? An alternative version might be: exiftool img_1.jpg | diff - <(exiftool img_2.jpg)

full command: for fn in xkcd*.png xkcd*.jpg; do; echo $fn; read xw xh <<<$(identify -format '%w %h' $fn); nn="$(echo $fn | sed 's/xkcd-\([0-9]\+\)-.*/\1/')"; wget -q -O xkcd-${nn}.json http://xkcd.com/$nn/info.0.json; tt="$(sed 's/.*"title": "\([^"]*\)", .*/\1/' xkcd-${nn}.json)"; at="$(sed 's/.*alt": "\(.*\)", .*/\1/' xkcd-${nn}.json)"; convert -background white -fill black -font /usr/share/fonts/truetype/freefont/FreeSansBold.ttf -pointsize 26 -size ${xw}x -gravity Center caption:"$tt" tt.png; convert -background '#FFF9BD' -border 1x1 -bordercolor black -fill black -font /usr/share/fonts/truetype/freefont/FreeSans.ttf -pointsize 16 -size $(($xw - 2))x -gravity Center caption:"$at" at.png; th=$(identify -format '%h' tt.png); ah=$(identify -format '%h' at.png); convert -size ${xw}x$(($xh+$th+$ah+5)) "xc:white" tt.png -geometry +0+0 -composite $fn -geometry +0+$th -composite at.png -geometry +0+$(($th+$xh+5)) -composite ${fn%\.*}_cmp.png; echo -e "$fn $nn $xw $xh $th $ah \n$tt \n$at\n"; done this assumes that all comics are saved as xkcd-[number]-[title].{png|jpg}. it will then download the title and alt-text, create pictures from them, and put everything together in a new png-file. it's not perfect, but it worked for nearly all my comics. it uses the xkcd-json-interface. though it's poorly written, it doesn't completely break on http://xkcd.com/859/

This is an "argument calculator" funktion. The precision is set to 4 and you can use dot (.) or comma (,) as decimal mark (which is great for german users with a comma on the numpad).

This opens a python command line. You can use math and random and float-division is enabled (without appending .0 to integers). I just don't know how to specify a standard precision.

reverse the sorting of ls to get the newest file: ls -1tr --group-directories-first /path/to/dir/ | tail -n 1 Problems: If there are no files in the directory you will get a directory or nothing.

the google-api gives you only one translation which is sometimes insufficent. this function gives you all translations, so you can choose which one fits best.

this command prints itself out. it doesn't need to be stored in a file and it isn't as easy as echo $BASH_COMMAND for information on quines see http://en.wikipedia.org/wiki/Quine_(computing)

i wanted to delete all duplicate lines from .bash_history and keep the order of the other lines. the command cat's the file and adds line numbers, then sorts by the second column. afterwards uniq omits repeated lines, but skips the first field (the line number). then it sorts by the line numbers and at the end cuts the numbers off.