Let me see if I can sort this out...
In the
Windows world, "there is basically
one way to do it." There is basically only
one operating-system environment that is possible, one type of hardware and so-on, and that environment has been rigorously defined by Microsoft. Software developers can therefore build "installers" that simply take care of the entire process of correctly installing software: there are not too many choices to make.
In the
Linux world, on the other hand, "
anything goes," and believe me I do mean,
"anything." The poor software developer does not even know what kind of CPU the target machine might be running, nor if so, in exactly what environment.
So here, in general, is how the developer solves the problem, in reverse order:
(0) Someone might have solved the following problem for you by creating a "package" that you can simply "install." If such a package exists, count your many lucky-stars and do that ... send a nice financial contribution to that kind good soul ... because, thanks to his or her
hard work, you can now avoid
all of the following steps...
(1) The final deliverable is probably going to be a
.tar.gz file, affectionately known as "a
tarball." This is a compressed file, much like a "zip." You'll extract the content of the file with a command like
tar xzvf filename.
(2) The first thing you'll be expected to do is to run a special script file, present in the tarball, called "configure." The magic is: first
cd into the directory that was created, then issue
./ configure.
Notice the leading period, slash, and space: they're all important.
(3) The
configure script will grind away for a while. Maybe a
long while, but consider just what it is doing. This script is
figuring out how to build this program on
your machine.
(4) When the configure-script has finished, it has built a "Makefile." So, all you need to do now is to type:
make. The computer will now
compile ... literally generate ... the actual program that will eventually run on your computer.
(5) After this process is finished, the routine command to be issued to install the software is:
make install.
(6) There are many variations. For instance, in the command above, "install" is what's called a
make target. The Makefile can define many "targets." Applications vary considerably in this regard.
(7) The
configure step can have many
options. Sometimes, the "voodoo" that must be applied at this step to get the intended result can be very spooky indeed. But it's all there for a reason: remember, the application is designed to install itself and to work properly in an environment that is, to the application designer,
completely unpredictable.
-----
If you ever wondered what a "distro" developer is actually doing, now you know. "Distros" decide upon a well-chosen set of "standard configurations" for what is actually a
huge collection of software, on a few "standard" machines, and they've figured out a way to devise a "known-good configuration" that works on them and then to
deploy it to you. They do their job amazingly well... well enough that you grouse when anything goes wrong, and well enough that folks like me are
stunned that it even works at all.
"Credit Where Credit Is Due" Dept.: One thing's for sure about Linux... it shows you just how
good Microsoft Corporation actually
is at "what they do." No matter what you do or don't feel about "Gates-Co," their technical achievement is
also "more amazing than (they want you to have to have) ever dreamed." In a world where "anything could go
catastrophically wrong at
any step..." it usually
doesn't. And
that, my friends and neighbors, just happens to be the product of amazingly hard work.
