On Mon 06/04/2020 10:47, Renaud Allard wrote: > Hello, > Here is a port of sniproxy 0.6.0 > > It does Name-based proxying of HTTPS without decrypting traffic. It's quite > lighter than some other alternatives when you only need this feature. > > Any comments? OK?
Upstream has a branch, which provides OpenBSD-specific bits: https://github.com/dlundquist/sniproxy/tree/openbsd. Did you know about this branch? If so, why are your patches better? Some comments below. Please note that I did not run tested this port. -COMMENT = Name-based proxying of HTTPS without decrypting traffic +COMMENT = name-based proxying of HTTPS without decrypting traffic Comments usually starts lower case. -DISTNAME = sniproxy -VERSION = 0.6.0 -PKGNAME = ${DISTNAME}-${VERSION} - GH_ACCOUNT = dlundquist GH_PROJECT = sniproxy -GH_TAGNAME = ${VERSION} +GH_TAGNAME = 0.6.0 GH_*-bits will take care of DISTNAME and PKGNAME for you. BUILD_DEPENDS = net/udns \ devel/libev \ devel/pcre \ - devel/gettext + devel/gettext,-tools 'make' errors out in the configure phase complaining about missing AC-*-goo, which is provided by gettext-tools. AUTOMAKE_VERSION = 1.16 -CONFIGURE_STYLE = gnu +CONFIGURE_STYLE = autoreconf +AUTORECONF = ./autogen.sh -do-gen: - cd ${WRKSRC}; ${AUTOCONF_ENV} ./autogen.sh Instead of defining a do-gen-phase use CONFIGURE_STYLE and AUTORECONF. -NO_TEST = Yes sniproxy provides test. I think it makes sense to have a look at the failing ones: ============================================================================ Testsuite summary for sniproxy 0.6.0 ============================================================================ # TOTAL: 22 # PASS: 14 # SKIP: 1 # XFAIL: 0 # FAIL: 7 # XPASS: 0 # ERROR: 0 Does it make sense to add a bit of explanation somewhere that /etc/sniproxy.conf is expected to be present? Once build, installed and provided a valid config, I could not get sniproxy to reload a new configuration using rcctl. Maybe set 'rc_reload=NO' in sniproxy.rc?
