Since the docs say that getting confused on first trying to use fpm may constitute a bug in the docs, I thought I'd email. Probably I'm just missing something...
What I want to do is to make deb files from a number of libraries and binaries that we build, mostly in C++ or C, to support our production systems. Some of these are the base libraries unavailable from apt, that we build from git and install once at machine creation, and only update occasionally. Some are the ones that we write ourselves and so deploy several times a day.
So I have a set of directories, and each one contains stuff that can be built with some set of commands, but let's say it's just ./autogen.sh && ./confgure && make && make install.
Now here's where I get confused. The fpm man page tells me to specify everything that needs to be installed and where it needs to go. Oftentimes "make -n install | wc -l" tells me it's got hundreds of lines of things it's going to do, so I'm not feeling like this has helped me too much. Conceptually, I expect to copy some libs, make some symlinks of various libs, run libtool, etc. But make install isn't trying to help me understand as it runs commands through sed to compute other commands.
I'm not sure exactly what I expected -- maybe that it attempted to capture the actions of make install via a chroot and then bundle those files?
Any pointers most welcome.
-- Jeff Abrahamson http://p27.eu/jeff/ http://transport-nantes.com/