> At that point, adding a legacy backend using fakechroot (or just chroot
> perhaps) would be a possibility and while I won't make any promises, I
> _might_ look at that later.
Hmm, I totally forgot that `chroot` alone wouldn't suffice, the system would
also need to have something like overlayfs and that's only available on Linux
AFAIK.
Overlayfs is used here to create CoW snapshots of the root filesystem, as
obviously making an actual copy for each test is too expensive and slow. The
old test suite works around this by cherry picking the core files and
directories to symlink into the per-test tree and lets fakechroot do its
trickery where it resolves those symlinks outside of the jail while making it
appear like a regular `chroot`. This kind of trickery is one of the reasons we
want to get rid of fakechroot as it's unpredictable and has weird side effects.
So it turns out the new test suite will only support Linux systems for the time
being. However, since it's made extensible, one can write a backend for
non-Linux systems that uses a CoW technology native to the given platform.
Adding a backend using fakechroot + symlinks (i.e. the old approach) would be
possible, too, but I don't think we want to go in that direction since removing
fakechroot is the main motivation for this PR (and the associated #1580).
--
Reply to this email directly or view it on GitHub:
https://github.com/rpm-software-management/rpm/pull/2559#issuecomment-1633321474
You are receiving this because you are subscribed to this thread.
Message ID: <rpm-software-management/rpm/pull/2559/c1633321...@github.com>
_______________________________________________
Rpm-maint mailing list
Rpm-maint@lists.rpm.org
http://lists.rpm.org/mailman/listinfo/rpm-maint