When building and installing from source, I redirect output to a file,
so that if anything goes wrong, i have the relevant info at hand.
So generally following these recipes (for tcsh - adjust for bash):
(0. verify download with sha1/md5/gpg sum/sig)
1. extract tarball:
a. i usually do a dry run (list/display, don't extract) first.
While most packages include a top-level directory, some do not,
in which i would create it and cd to it before extracting.
tar ztvf fileName.tar.gz | tail # or |less, etc.
b. extract: tar zxf fileName.tar.gz # with gzip
tar jxf fileName.tar.bz2 # with bzip2
Note that the 'f' probably needs to be last in the concatenated
tar options such as "zxvf", so as to be followed immediately
by the file name.
2. go into the directory - cd fileName/
3. a. configure >& config.out
grep -i error config.out
b. make >& make.log
grep -i error make.log
c. Some packages include a test suite which you can optionally run
with "make test" or "make check"
d. make -n install | less
for the paranoid, do a sanity check on the install script, before
doing anything as root (we did all the preceding as ordinary user,
right?) - review 'rm' statements and so forth.
e. sudo make install >& install.log
grep -i error install.log
f. as far as "make clean", that will *clean* out the derived files,
that is the ones generated from the original packaged ones,
thus not wasting as much disk space.
You could cd ..; rm -rf fileName (carefully!) and clean out more space.
I'd wait a week before doing either one, unless disk space is critical,
so see the successful functioning of the program for a week, before
wiping out the record of building it.
4. as far as being in your path, most packages install to /usr/local/bin/
and the install step (3e) puts it there.
HTH
Randall
