Shell Programming and Scripting

Hi everybody,
I have written a script that parses a file line by line.

I have used 2 methods.
1 using read function and the other using line function.

I timed both scripts and what I have observed is that the one with read executed atleast 100 times faster than the one scripted with line function.

Can you throw any light on my observation?

How do you explain the difference in efficiencies?

this is like your 3rd or 4th question about the differance between 2 commands.

something says this is not a general curiousity type of questions.

i suggest carefully reading the man pages and do some real thinking.

if you want to know about various commands and how they execute then looking into the man page for truss and what not.

Well this is one of the most important concepts in shell programming, so I'm going to answer it. "line" is a program. "read" is a builtin. This is the key to lightning fast shell scripts.

ksh and bash and some modern vershions of sh have the "read" command. When these shells encounter a "read" command, they issue a read system call..

But line is a program. The shells must scan the directories in your PATH variable. If they find the program, then they can run it. So first they fork() to create a new process. The new process then will exec() the "line" program. Now the line program can do a read(). Once it has a line, it has to write the line to stdout. Now it can exit(). Meanwhile, the original shell process has issued a wait(). With the death of its child process, it must now wake up. Now it can read() the line.

This (rough) sequence of events happens every time you invoke cut or sed or awk to massage a variable. And every time that you use expr to do a calculation.

Fight hard to avoid external programs. Especially in a loop.

Another example of using builtins... Counting time