To clarify, I am wondering how you can install RPM on a system at the installation time. Yes, I know that to install you use "rpm -i", but that is not the question.
My point is that when you install a new CentOS for instance, we will
- Boot on a minimal system
- Start the graphical environment
- Configure the UI parameter for the installation process
- Setup the disk (partitioning, formatting)
- List the packages you are expecting to install
- Install those packages
- Create user accounts and stuff
And all of this will be managed by Anaconda.
My point is that I know how you create the partitions (fdisk, parted, etc...), I know how to format them (mkfs, ...), then you can create the filesystem tree (mkdir, ...), but then how would you be able to deploy the RPM, since there is no RPMDB, and no RPM utility deployed.
I guess that a chroot might be involved so that you can isolate your new system, then we're on our own. Needless to say when checking about RPM installation, I just get basic RPM installation commands. Any tracks to provide on how we would be able to do that ?