Re: bash vs. python scripts - which one is better?

["Douglas A. Tutty" <dtutty@porchlight.ca>]

On Sun, Aug 19, 2007 at 02:40:19PM -0700, Steve Lamb wrote:
> David Brodbeck wrote:
> > In particular, Bash's "test" (aka "[")
> >operator has pitfalls.  Testing for an empty variable, for example, is 
> >awkward.  If you do:
> 
> >if [ $foo == "" ]
Yeah, and the spaces between the [ $ and the " ] are critical too; I
just forget in what way.
> 
>     Yeah, prefer:
> 
> if not foo:
>     do something
> 
> >-- there are few languages where it's quite so easy to test conditions 
> >like, 'is this a directory?'.
> 
>     Bash undoubtedly is more concise than this but I'd contend it is no 
>     easier.
> 
> import os
> if os.path.isdir(somedir):
>     print "It's a dir, Jim!"

Best of all, at 2:00 a.m. a year from now, its perfectly clear what
 if os.path.isdir(somedir): 

 means.

> 
> >Bash is great if you need to glue together a bunch of existing utilities 
> >to do something.  You can pipe output directly from one utility to 
> >another with an ease that doesn't exist in most other languages, which 
> >tend to make it complicated to launch other processes.  Doing complex 
> >tasks with a series of simple utilities is sort of what Unix is all about.

Right, but debugging a long piece of plumbing can take a while.  

> 
>     I contend that is more a was than an is.  Shell filled a niche years 
>     ago that now has largely shrunk to special cases.  So far in this thread 
> there have been two example total of where a shell script might be better 
> for more than 2 minutes over a proper scripting language.  Generally by the 
> time any shell script starts piping 2-3 things I just convert it to Python 
> and get it done far easier and far faster.  Mainly because I don't have to 
> jump through the hoops the warts above present.
> 
> >Besides, until operating systems start having init scripts written in 
> >Perl or Python, being able to write shell scripts is going to be an 
> >essential system administration skill. ;)

Anything after /usr is mounted (perhaps runlevel 2) could be in python.
The init scripts for rcS.d should be carefully written to be
understandable by non-bash gurus.

Check out the very first initscript, S01glibc.sh (exerpts below).  It
may as well be written in assembler for all I can understand how it
compares kernel versions.

Doug.



if [ "`uname -s`" = Linux ]; then
    # glibc kernel version check: KERNEL_VERSION_CHECK
kernel_compare_versions () {
    verA=$(($(echo "$1" | sed 's/\([0-9]*\)\.\([0-9]*\)\.\([0-9]*\).*/\1 \* 10000 + \2 \* 100 + \3/')))
    verB=$(($(echo "$3" | sed 's/\([0-9]*\)\.\([0-9]*\)\.\([0-9]*\).*/\1 \* 10000 + \2 \* 100 + \3/')))
    
    test $verA -$2 $verB
}

exit_check () {
    sleep 5
    exit 1
}

    # Test to make sure z < 255, in x.y.z-n form of kernel version
    # Also make sure we don't trip on x.y.zFOO-n form
    #kernel_rev=$(uname -r | tr -- - . | cut -d. -f3 | tr -d '[:alpha:]')
    kernel_rev=$(uname -r | sed 's/\([0-9]*\.[0-9]*\.\)\([0-9]*\)\(.*\)/\2/')
    if [ "$kernel_rev" -ge 255 ]