openssl-1.0

Category: local-install (rebuilt only so archlinuxaba/the cluster gets it too; not maintained/adapted/owned by us — see arch/scripts/install_only_packages). Compat OpenSSL 1.0.2 build (libssl.so.1.0.0/libcrypto.so.1.0.0, headers/binary under openssl-1.0-namespaced paths) for software that still needs the old ABI, e.g. python2 (see memory/python2.md).

Build notes (2026-08-19): extended to i686/pentium4

• Local checkout was byte-identical to the real upstream AUR PKGBUILD (diff against cgit/aur.git/plain/PKGBUILD?h=openssl-1.0 showed no differences) but declared arch=('x86_64') only, even though build() already had a working i686 branch (openssltarget='linux-elf') baked in by upstream itself – just never exposed via arch=().
• Added pentium4 to the same elif branch as i686 (same 32-bit x86 Configure target, linux-elf) and extended arch=('x86_64' 'i686' 'pentium4'). Built and published cleanly on the first try, all three arches.
• check()’s OCSP test prints several Verification: FAILED / === INVALID SIGNATURE ... === lines during make test – these are the test suite’s own expected-negative-result cases (intentionally malformed OCSP responses/certs), not real failures. Confirmed by all six packages (openssl-1.0 + openssl-1.0-debug per arch) actually getting UPLOADED; a real check() failure would have printed === <arch> FAILED === and skipped the upload.
• pkgrel 7 (upstream) -> 7.1 (local-install, X.Y scheme – the arch extension is a local deviation, not an upstream release).
• Found and cleaned up stale untracked build artifacts (old .pkg.tar.zst, downloaded tarball, pkg//src/ dirs) left over from an earlier, apparently interrupted build attempt before this one – harmless, just noting in case a similar leftover shows up again.