A number of variables were previously removed from PLIST_SUB for being too generic are now present in PLIST_SUB after r420328, with affects such as: Error: Orphaned: %%LIB32DIR%%/python%%XYDOT%%/idle%%LIB32DIR%%/tabbedpages.py Looking into Mk/bsd.port.mk, there appears to be at least one reference to an undefined variable on line 1609: PLIST_SUB_SED_tmp2= ${PLIST_SUB_tmp1: ... I presume this should be PLIST_SUB_SED_tmp1 After correcting this, LIB32DIR is still present in PLIST_SUB Contents of relevant variables: irc/znc] make -V PLIST_SUB_SED_tmp1 CHARSET="@comment " CYRUS="@comment " DEBUG="@comment " DOCS="@comment " NO_IPV6="@comment " NO_OPENSSL="@comment " NO_PERL="@comment " PYTHON="@comment " TCL="@comment " OSREL=12.0 LOCALBASE=/usr/local RESETPREFIX=/usr/local PORTDOCS="@comment " PORTEXAMPLES="@comment " LIB32DIR=lib PERL_VERSION=5.22.3 PERL_VER=5.22 PERL5_MAN1=lib/perl5/site_perl/man/man1 PERL5_MAN3=lib/perl5/site_perl/man/man3 SITE_PERL=lib/perl5/site_perl SITE_ARCH=lib/perl5/site_perl/mach/5.22 DOCSDIR="share/doc/znc" EXAMPLESDIR="share/examples/znc" DATADIR="share/znc" WWWDIR="www/znc" ETCDIR="etc/znc" No generic variables here: irc/znc] make -V PLIST_SUB_SED_tmp2 PERL_VERSION=5.22.3 PERL_VER=5.22 PERL5_MAN1=lib/perl5/site_perl/man/man1 PERL5_MAN3=lib/perl5/site_perl/man/man3 SITE_PERL=lib/perl5/site_perl SITE_ARCH=lib/perl5/site_perl/mach/5.22 DOCSDIR="share/doc/znc" EXAMPLESDIR="share/examples/znc" DATADIR="share/znc" WWWDIR="www/znc" ETCDIR="etc/znc" Generic variables here: (_tmp3 uses _tmp1, not _tmp2, not sure if intended) make -V PLIST_SUB_SED_tmp3 CHARSET="@comment " CYRUS="@comment " DEBUG="@comment " DOCS="@comment " NO_IPV6="@comment " NO_OPENSSL="@comment " NO_PERL="@comment " PYTHON="@comment " TCL="@comment " OSREL=12.0 LOCALBASE=/usr/local RESETPREFIX=/usr/local PORTDOCS="@comment " PORTEXAMPLES="@comment " LIB32DIR=lib PERL_VERSION=5.22.3 PERL_VER=5.22 PERL5_MAN1=lib/perl5/site_perl/man/man1 PERL5_MAN3=lib/perl5/site_perl/man/man3 SITE_PERL=lib/perl5/site_perl SITE_ARCH=lib/perl5/site_perl/mach/5.22 DOCSDIR="share/doc/znc" EXAMPLESDIR="share/examples/znc" DATADIR="share/znc" WWWDIR="www/znc" ETCDIR="etc/znc" Generic variables here in final PLIST_SUB: irc/znc] make -V PLIST_SUB CHARSET="@comment " NO_CHARSET="" CYRUS="@comment " NO_CYRUS="" DEBUG="@comment " NO_DEBUG="" DOCS="@comment " NO_DOCS="" IPV6="" NO_IPV6="@comment " OPENSSL="" NO_OPENSSL="@comment " PERL="" NO_PERL="@comment " PYTHON="@comment " NO_PYTHON="" TCL="@comment " NO_TCL="" OSREL=12.0 PREFIX=%D LOCALBASE=/usr/local RESETPREFIX=/usr/local PORTDOCS="@comment " PORTEXAMPLES="@comment " LIB32DIR=lib PERL_VERSION=5.22.3 PERL_VER=5.22 PERL5_MAN1=lib/perl5/site_perl/man/man1 PERL5_MAN3=lib/perl5/site_perl/man/man3 SITE_PERL=lib/perl5/site_perl SITE_ARCH=lib/perl5/site_perl/mach/5.22 DOCSDIR="share/doc/znc" EXAMPLESDIR="share/examples/znc" DATADIR="share/znc" WWWDIR="www/znc" ETCDIR="etc/znc"
I just realised the wrong variable (PLIST_SUB) was used to check the presence of generic variables rather than PLIST_SUB_SED, which was throwing off my tests. Patch (phabricator review) attached fixing two variable references. For lang/python36, the generic variables are no longer present. Custom variables added by the port remain present lang/python36] make -V PLIST_SUB_SED s!3\.6!%%XYDOT%%!g; s!3\.6\.0a4!%%XYZDOT%%!g; s!share/doc/python!%%DOCSDIR%%!g; s!share/examples/python!%%EXAMPLESDIR%%!g; s!share/python!%%DATADIR%%!g; s!www/python!%%WWWDIR%%!g; s!etc/python!%%ETCDIR%%!g;
Ah, yes, once again, it's what the lack of review does.
A commit references this bug: Author: mat Date: Fri Aug 19 13:09:29 UTC 2016 New revision: 420461 URL: https://svnweb.freebsd.org/changeset/ports/420461 Log: Fixup the PLIST_SUB_SED creation. PR: 211995 Reported by: koobs Sponsored by: Absolight Changes: head/Mk/bsd.port.mk