I got the following from
http://www.linuxforums.org/forum/viewtopic.php?p=105598. A useful little
note. Thanks to the author.
Well, installing from an RPM doesn’t compile the program. The RPM file
contains the already-compiled program and just copies it into place. It
also saves info in a database about what files belong to what package
and where they are installed.
Compiling a program from source, on the other hand, has the advantage
that the program will be optimized for your hardware and system. For
example, if you have an AMD Athlon, the compiler can use some
Athlon-specific instructions to make your program faster, while the
pre-compiled programs in RPM files must be compatible with all
processors, so they are always compiled with the 386 as a template.
Also, if you have some additional programs installed on your system,
like Kerberos or anything, a program compiled from source can take
advantage of that, while a program in a RPM file always uses as little
extra functionality as possible, to be able to install on all systems,
regardless of whether that system has the extra functionality installed
RPM does have the not-to-be-denied advantage that it stores info about
where all the files have been installed to, though. That both eases
uninstallation and makes it possible to query unknown files. For
example, if you find a file called /usr/bin/onsgmls and want to know
what it is, you can run “rpm -qf /usr/bin/onsgmls” to find out that it
belongs to that “openjade” package. Then you can run “rpm -qi openjade”
to find out what openjade is. If you find out that maybe you don’t want
the openjade package, you can easily uninstall it by running “rpm -e
openjade”. Or anything… see “man rpm” for more options and possibilities.
Fedora Core 1 and 2 also ship with the yum utility, which allows you to
do automatic RPM installations. For example, if you wanted mplayer, you
could just run “yum install mplayer” (probably – I haven’t tried,
but…), and it would automatically download and install it for you.