Configure, make, make install

[miguelpdl]

This is an enhancement feature request.

The current process to build the IRATI Stack is

sudo ./configure --prefix ~/install
sudo make install

But the process should be

./configure --prefix ~/install
make
sudo make install

Now I (think) I know why the build is done like this. The configure actually builds and installs librina so that when it comes to rinad it has librina in place. But is there another way to complete this in a neat way.

This should be investigated.

[TRIA]

A good reason for separating make from make install is that if the build can be done under the user's uid, the owner of the source and build tree need not be super-user, since ./configure and make only write into the user's directories. This is handy, for example, to allow scp'ing source files around without having to enable super-user ssh login (a bad idea). This approach is also useful for doing private testing prior to public installation. This is less important for things like kernel modules, but useful for testing other code while preserving a working version in public places.

So when fixing this, ideally, the ./configure and make steps should not write anyplace that normal users don't have write permissions for. This requires the generated makefiles to reference the build-directory versions of static libraries and .so's for all built binary programs (it would be incorrect to use the public installed versions, which might not work with the new code being built).

Therefore, make should build kernel modules that are ready to install, and libraries and binaries that work in-place in the build tree, and make install will first rebuild libraries to correct library location dependencies (only if necessary) and then install the libraries in the final destination in public place(s), then relink all executables that depend on them to depend on the public location of the libraries.