UNIX for Dummies Questions & Answers

Hello,

This would be very basic question to most of you, but for me as a newbie it is an un-know.

I would like to load a configuration file into memory before executing a specific command. For example: I have a config file that holds all the database connectivity information and I'd like to load this specific db config file in order to execute a command:
config file will have:

Code:

export user=xyz
export password=abc
export database=123

Code:

mysql -u $user -password $password -db $database

I need the values correctly read from the config file that I will load before executing the above command.
Help is much appreciated.

I'd rather not export password's....

Make a script that looks like:

Code:

#!/bin/bash
source /path/to/password-file
mysql -u $user -password $password -db $database

Password-file

Code:

user=xyz
password=abc
database=123

This way, the variable 'password' is also emptied when the script ends.

Hope this helps

Hello Sea,

Thanks for the quick turn around, that's one way of doing it which I am aware of. But I need something that I can keep loading dynamically just before executing the command. Password was just an example I provided, I have other variables that I need to load in other scenarios. I need something that I need to do at command line.
like:

Code:

. ./.myconfigFile
mysql -u $user -password $password -db $database

do we have something like that in linux, to load configuration at command line dynamically, apart from the one way I have posted in this reply.

This is exactly how it's done. Why do you want to change it?

Your request and the example code of both of us is all the same.
I'm not understanding your question.

Apart from what sea and RudiC already said (and which i can wholeheartedly agree with) here are some other things to consider:

Code:

export variable=value

is a quite common malpractice but still a malpractice. "export" is a command (in fact a "reserved word") and a variable declaration is a command. You cannot and should not mix that up. Do it this way:

Code:

variable=value
export variable

or, if you insist on one line:

Code:

variable=value ; export variable

btw., the effect of "export" on a variable is lasting. You do not need the also oftenly seen:

Code:

variable=value ; export variable
variable=newvalue ; export variable
variable=othervalue ; export variable
....

Once a variable is exported, it stays exported ("exported" means: its current value is copied to the environment of any child process it creates) and if you change its value then this new value instead of the old one would be copied to a newly inherited child.

If you "dot-execute" a script like this:

Code:

. /some/script.sh

this means that it is executed in the current envrronment instead of creating a new one at the start which is closed at the end (which is the default). This is why changes to the environment are lasting. On the other hand you can use any shell device there is, including control structures. You could accept a parameter and set your environment accordingly, like in the following sketch:

Code:

#! /bin/ksh

if [ $# -ne 0 ] ; then
     print -u2 - "Error: provide a parameter!"
     exit 2
fi

case "$1" in
     "one")
          user=userone
          database=dbone
          password=pwone
          ;;

     "two")
          user=usertwo
          database=dbtwo
          password=pwtwo
          ;;

     "three")
          user=userthree
          database=dbthree
          password=pwthree
          ;;

     *)
          print -u2 - "Error: unkown value, use only \"one\", \"two\" or \"three\","
          exit 1
          ;;

esac

exit 0

which you would call in this way:

Code:

. /path/to/script one
. /path/to/script two
. /path/to/script three

I hope this helps.

bakunin