On Mon, Jul 10, 2000 at 12:41:49PM +0200, Bernhard R. Link wrote: > I had to change a little bit on a system boild on top of an old suse. > It was terrible. The script called a script, which called onother, which > called the first with different paramters which called aanother. After the > 15th (!!) script I got lost. (Perhaps earlier,but I didn't notice before). > So I compared the file it got it input from with the man-page of the > command to be invoked and guessed what it should do. (It took me 4 hours, > before I did so, because I had to search the file manually, it wasn't > mentioned anywhere in the first 15 scriptfiles). this is precisly why i don't want any of this function nonsense. the most an initscript should source is the iniscript.conf file like so: VAR=default if [ -f /etc/config.d/initscript.conf ] ; then . /etc/config.d/initscript.conf fi which is what joey proposed i think. anything more then that is uncessary obfuscation. > What I kept was a nightmare against any script-system, one calling each > other. One file mit Variables may be OK for easy changes, so that dpkg > can overide the main-sript, because it is not changed. But the file itself > should almost always be a conffile. > > > And people don't have to write everything from scratch. They have s-s-d. > It is the debian way of making it clear and simple. /etc/init.d/skeleton. for most daemons you only need to fill in the blanks so to speak in a copy of this file and you have a consistent, understandable initscript. no need for any of this ball of funtions squirelled away somewhere. -- Ethan Benson http://www.alaska.net/~erbenson/
Attachment:
pgpw_Jyn1ZRMP.pgp
Description: PGP signature