-
Bug
-
Resolution: Fixed
-
Important (Ebuild)
-
None
-
None
-
None
-
random environment variables can break openssl build.
>>> Preparing source in /var/tmp/portage/dev-libs/openssl-1.1.1q/work/openssl-1.1.1q ... * Applying openssl-1.1.0j-parallel_install_fix.patch ... [ ok ] * Applying openssl-1.1.1i-riscv32.patch ... [ ok ] * Applying openssl-3.0.5-test-memcmp.patch ... [ ok ] Operating system: x86_64-whatever-Linux This system (Linux) is not supported. See file INSTALL for details.
I found a really corner case issue. Currently the `config --test-sanity` command execute on src_prepare section uses env variables.
Between these variables it uses these:
[ "$MACHINE" ] || MACHINE=`(uname -m) 2>/dev/null` || MACHINE="unknown" [ "$RELEASE" ] || RELEASE=`(uname -r) 2>/dev/null` || RELEASE="unknown" [ "$SYSTEM" ] || SYSTEM=`(uname -s) 2>/dev/null` || SYSTEM="unknown" [ "$BUILD" ] || VERSION=`(uname -v) 2>/dev/null` || VERSION="unknown"
But if the env variable RELEASE was already set (in my case set to `next`), this break the OS detection.
Could be i good idea on src_prepare section run
unset MACHINE RELEASE SYSTEM BUILD
before run the sanity check.