Log: Difference between revisions

From genomewiki
Jump to navigationJump to search
No edit summary
No edit summary
 
(5 intermediate revisions by the same user not shown)
Line 1: Line 1:
This was inspired by a [http://www.kuro5hin.org/story/2005/12/28/223217/93 post on kuro5hin] some time ago, questions from wet-lab collegues ("you don't keep protocols?????"), own negative experiences ("How the heck did I create this damn bed-file three days ago?") and also by the makeDbxxxx files in the source tree. Of course, makefiles are better than shell scripts, but often you're just hacking around and don't want to bother with escaping those $s, tab-characters, etc. (It would be nice to know what kind of scripts the UCSC people use to for their makeDbxxx protocols...) (We write our shell commands in the makeDb/*.txt files first, then cut and paste those commands to the command line.  If they are incorrect, they are corrected in the *.txt file, then cut and paste again to the command line until they work right. The *.txt file is open in an editor until a sequence of operations is completed. --Hiram)
I have found a very easy way to keep logfiles of the commands that you executed in your shell to generate a certain annotation file for genome. It was inspired by a [http://www.kuro5hin.org/story/2005/12/28/223217/93 post on kuro5hin] some time ago, Of course, makefiles are better than logfiles, but often you're just hacking around and don't want to bother with escaping those $s and tab-characters.


Add these two lines to your .bashrc:
The people at UCSC keep their logfiles in an editor: "We write our shell commands in the makeDb/*.txt files first, then cut and paste those commands to the command line.  If they are incorrect, they are corrected in the *.txt file, then cut and paste again to the command line until they work right. The *.txt file is open in an editor until a sequence of operations is completed. --Hiram"
  alias logadd='history 2 | head -n 1 | cut -d" " -f4- >> log'
  alias logmenu=source ~/usr/bin/scripts/logmenu


(Please adjust the path ~/usr/bin/scripts here to the directory where you keep your scripts)
I prefer my way as it saves me the copy-pasting. The following script allows you to write the last command to a logfile and later run selected commands from your log.


When you type '''logadd''' now, the last command you typed will be appended to a file called "log".  
Add these two lines to your .bashrc:
  export HISTIGNORE="logadd:log"
  alias logadd="history -p \!\! >> log.txt"
  alias log="vim log.txt"
When you type '''logadd''' now, the last command you typed will be appended to a file called "log.txt".  
So its a good habit to use '''logadd''' after you have found just the right combination of blat, pslSelect, overlapSelect and faPolyASizes, to be able to track down those nasty mistakes 6 months later.
So its a good habit to use '''logadd''' after you have found just the right combination of blat, pslSelect, overlapSelect and faPolyASizes, to be able to track down those nasty mistakes 6 months later.
Then save the following file under the name '''logmenu''' in the directory where you keep your scripts:
<pre>
# this scripts needs alias logmenu=source ~/usr/bin/scripts/logmenu
# in your bashrc otherwise the cmds will not find the way into your history
#!/bin/bash
if [ "$1" == "-h" -o -z $1 ]; then
    echo logmenu: display the file \"log\" as a menu and let the user choose a command to execute
    echo uses the program \"dialog\"
fi
echo -n dialog --menu logfile 24 70 18\  > /tmp/menu
cat log | tr -d \' | gawk "{ ORS=\" \"; print NR, \"\'\" \$0 \"\'\" } " >> /tmp/menu
. /tmp/menu 2> /tmp/menuresult
line=`cat /tmp/menuresult`
echo $line
head -n $line log | tail -n 1 > /tmp/line
history -r /tmp/line
. /tmp/line
rm -f /tmp/line /tmp/menuresult /tmp/menu
</pre>
Now, when you type '''logmenu''' somewhere, a nice menu will pop up, you can select a command, which will then be run and also put into your history, so you can press the up-arrow-key, modify it, run it again, etc.


[[Category:User Developed Scripts]]
[[Category:User Developed Scripts]]

Latest revision as of 17:49, 15 February 2015

I have found a very easy way to keep logfiles of the commands that you executed in your shell to generate a certain annotation file for genome. It was inspired by a post on kuro5hin some time ago, Of course, makefiles are better than logfiles, but often you're just hacking around and don't want to bother with escaping those $s and tab-characters.

The people at UCSC keep their logfiles in an editor: "We write our shell commands in the makeDb/*.txt files first, then cut and paste those commands to the command line. If they are incorrect, they are corrected in the *.txt file, then cut and paste again to the command line until they work right. The *.txt file is open in an editor until a sequence of operations is completed. --Hiram"

I prefer my way as it saves me the copy-pasting. The following script allows you to write the last command to a logfile and later run selected commands from your log.

Add these two lines to your .bashrc: 

 export HISTIGNORE="logadd:log"
 alias logadd="history -p \!\! >> log.txt"
 alias log="vim log.txt"

When you type logadd now, the last command you typed will be appended to a file called "log.txt". So its a good habit to use logadd after you have found just the right combination of blat, pslSelect, overlapSelect and faPolyASizes, to be able to track down those nasty mistakes 6 months later.

Retrieved from "https://genomewiki.ucsc.edu/index.php?title=Log&oldid=22436"