currently we blacklist certain phases (which is largely based on EAPI=0 and
blocking src_*) for enew{user,group}. moving forward, ferringb suggested we
invert this into a whitelist of allowed phases.
afaict, the blacklisting + dev documentation has done a good job of
restricting calls to three places: pkg_{setup,preinst,postinst}. so inverting
the logic should largely be safe. on the off chance it isn't, i think letting
the ebuild `die` and getting it fixed up via bug reports is acceptable (i
grepped through the tree a bit and looked sane).
moving beyond that, i'd like to also ban pkg_postinst usage. the trouble with
using this phase is that `die` isn't fatal because the package has already
been merged to $ROOT, so there's no going back. and while logically i can see
that people might prefer pkg_postinst vs pkg_preinst (no point in creating a
user/group if the pkg isn't actually yet merged), for all real world usage,
there's no need to delay it, and it makes the ebuilds a bit more robust as
errors get caught before things get merged to $ROOT. i believe the
documentation has always recommended pkg_setup and pkg_preinst anyways.
moving even beyond that, i'd like to update the documentation to push people
to prefer pkg_preinst. if your package doesn't require the user/group to be
available at compile/install time, then there's no need to use pkg_setup.
this improves the use case of attempting to install a package from source,
things failing for whatever reason, and then the pkg is never actually
installed, but the user/group is left behind.
any feedback before i implement ?
-mike
signature.asc
Description: This is a digitally signed message part.