regess/infra-unittests: add test for unknown configure option

(rillig)

devel/lmdbg: fix typo in REPLACE_SH

(rillig)

x11/kitty: remove nonexistent file from REPLACE_PYTHON

(rillig)

cad/freehdl: remove nonexistent file from REPLACE_PERL

(rillig)

benchmarks/fio: remove nonexistent file from REPLACE_PYTHON

(rillig)

editors/emacs25: don't try to fix nonexistent grep-changelog

(rillig)

benchmarks/dnsperf: remove nonexistent file from REPLACE_SH

(rillig)

security/dehydrated: remove nonexistent file from REPLACE_BASH

(rillig)

games/crossfire-server: remove nonexistent file from REPLACE_PERL

(rillig)

net/argus: remove no-op REPLACE_PERL

(rillig)

emulators/fs-uae: document suspicious array[char]

(rillig)

devel/bugzilla: remove no-op file from SUBST_FILES

(rillig)

mk/configure: remove outdated documentation

(rillig)

mk/configure: add variable GNU_CONFIGURE_QUIET

This allows bulk builds or other setups to see interesting error
messages, without them being hidden in the noise of "checking ...".

(rillig)

net/couriertcpd: remove ipv6 option

The configure script does not know --with-ipv6 anymore.

(rillig)

mk/pkgformat/metadata.mk: fix relative paths in PATCHDIR and FILESDIR

Before, relative paths had been stored as-is.  This affected those
packages that defined PATCHDIR or FILESDIR as relative directory instead
of prefixing it with ${.CURDIR}.

Since there are already several other paths that are interpreted relative
to the package directory (CONFLICTS, DEPENDS), allow PATCHDIR and
FILESDIR to be specified as relative paths, too.  This makes the package
Makefiles a bit shorter.

(rillig)

cross/ppc-morphos-gcc9: skip portability check for libgo

(rillig)

games/freeciv-share: remove redundant SUBST block

Since common/Makefile.in is already patched, common/Makefile doesn't need
to.

(rillig)

games/freeciv-*: remove unknown configure option

(rillig)

x11/vte: remove unknown configure option

(rillig)

devel/pkg-config: suppress USE_TOOLS+=perl option

(rillig)

mk/configure/gnu-configure.mk: disable built-in option check

The rewritten check for unknown configure options already checks for
options that are not defined anywhere.  After that, the configure scripts
don't need to check for these options anymore since all remaining options
are known to the package, in at least one of its configure scripts.

(rillig)

math/maxima: remove nonexistent file from REPLACE_SH

(rillig)

graphics/cinepaint: remove unknown configure option

(rillig)

misc/celestia: remove unknown configure option

(rillig)

comms/asterisk15: remove unknown configure option --with-ltdl

This option has been removed in 2018, see ChangeLog.

(rillig)

games/asc: remove unknown configure option --with-wx-version

This is needed to build the package with GNU_CONFIGURE_STRICT=yes.

(rillig)

mk/configure: completely rewrite check for unknown configure options

The previous implementation could not reliably detect outdated configure
options.  This was apparent in devel/gettext-tools, where the option
--with-included-libcroco had become unknown between May 2019 and May
2020, but the check was not run.

The behavior is the same in the pkgsrc default configuration.  Only if
GNU_CONFIGURE_STRICT=yes, the new check is activated and will make
packages fail that previously succeeded to build.  Since that variable is
not widely known, there won't be much sudden breakage, if any.

(rillig)

regress/tools-platform: add test for "$@" and ksh

(rillig)

net/aria2: remove unknown configure options

(rillig)

www/ap-auth-kerb: remove unknown configure option

This allows the package to be built with GNU_CONFIGURE_STRICT=yes.

Since other Apache modules may still need this option, use an opt-out
here.

(rillig)

mail/amavisd-milter: remove unknown configure options

(rillig)

lang/algol68g: fix outdated configure options

Found by GNU_CONFIGURE_STRICT.

(rillig)

graphics/Coin: fix typo in configure option

The option --enable-pthreads was unknown, as a build with
GNU_CONFIGURE_STRICT=yes showed.  Fixing the typo showed exactly the same
configure output as before, which is probably due to the autodetection.

(rillig)

mail/elm-me: document harmless array[char]

(rillig)

devel/ExmanIDE: remove no-op file from SUBST_FILES

(rillig)

mark packages that fail with -Werror=char-subscripts

These packages are susceptible to bugs when confronted with non-ASCII
characters.

See https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94182.

It takes some time to analyze and fix these individually, therefore they
are only marked as "needs work".

(rillig)

chat/centerim: document that the package is array[char]-safe

(rillig)

sysutils/u-boot: remove no-op SUBST block

Line 1 already contains python, not python3.

(rillig)

sysutils/xentools411: clean up SUBST blocks

Several files have switched from /usr/bin/perl to /usr/bin/env perl.

(rillig)

sysutils/xentools42: remove no-op file from SUBST_FILES

(rillig)

regress/infra-unittests: demonstrate relative paths in +BUILD_VERSION

Pointed out by wiz@.

This occured in math/libixion/Makefile.common until 2020-05-19, and still
occurs in math/xyconvert/Makefile.  In all other packages, PKGDIR is
prefixed with  ${.CURDIR} and is thus an absolute path.

It should not be necessary to always specify PATCHDIR as an absolute
path, and the code in mk/pkgformat/pkg/metadata.mk seems to be the only
place where relative paths are handled wrong.

(rillig)

regress/infra-unittests: remove unnecessary tmpdir handling

The test infrastructure from test.subr already takes care of setting up a
temporary directory.

(rillig)

regress/plus_build_version: the bug has been fixed in January 2020

(rillig)

doc: Updated pkgtools/pkglint to 20.1.7

(rillig)

pkgtools/pkglint: update to 20.1.7

Changes since 20.1.6:

Versioned Python dependencies may end with :test or :tool, as the code in
lang/python/versioned_dependencies.mk says.

(rillig)

devel/bzr-gtk: remove nonexistent file from REPLACE_PYTHON

(rillig)

mk/configure/replace-interpreter.mk: silently skip broken symlinks

Like directories, they probably come from shell globs.

Seen in misc/byobu.

(rillig)

benchmarks/bonnie++: clean up REPLACE_PERL

bon_csv2html has been rewritten in C++.

(rillig)

graphics/inkscape: clean up REPLACE_PYTHON

Some of the files don't exist anymore, the others don't need it.

(rillig)

editors/zim: clean up duplicate paths in SUBST_FILES

(rillig)

devel/swig2: suppress USE_TOOLS+=perl warning

(rillig)

x11/qt4-qdbus: suppress USE_TOOLS+=perl warning

(rillig)

databases/postgresql12-pltcl: suppress USE_TOOLS+=perl warning

(rillig)

databases/postgresql12-plpython: suppress USE_TOOLS+=perl warning

(rillig)

databases/postgresql12-contrib: suppress USE_TOOLS+=perl warning

(rillig)

devel/libthrift: suppress USE_TOOLS+=perl warning

(rillig)

lang/gcc9: fix build with -Werror=char-subscripts

(rillig)

devel/quilt: clean up SUBST blocks

(rillig)

x11/qt4-tools: suppress USE_TOOLS+=perl warning

(rillig)

cross/binutils-mips-current: remove unknown configure option

(rillig)

converters/libetonyek: fix build on NetBSD-8.0-x86_64

(rillig)

textproc/redland: disable check for unknown configure options

"bmake show-unknown-configure-options" said that everything is ok.

(rillig)

textproc/redland: remove unknown configure options

(rillig)

misc/libreoffice: remove no-op configure.ac from SUBST_FILES

It doesn't contain any placeholders of the form @VAR@.

(rillig)

lang/gcc49: build even with -Werror=char-subscripts

(rillig)

lang/gcc48: build even with -Werror=char-subscripts

(rillig)

lang/gcc10: fix patch checksums to include the bug report URL

(rillig)

lang/gcc10: fix build with -Werror=char-subscripts

(rillig)

cad/ghdl: skip portability check

(rillig)

devel/gdb: disable check for unknown configure options

(rillig)

regress/infra-unittests: add test for variable precedence

(rillig)

pkgtools/pkgin: allow no-op SUBST block for paths

In a pkgsrc setup with default paths, the simple substitutions are all
no-ops.  The last two substitutions only apply to Makefile.in, which
means that they are no-ops for the other files.  Since the whole sed
command is not considered an identity substitution, an explicit
SUBST_NOOP_OK is needed here.

https://mail-index.netbsd.org/pkgsrc-users/2020/05/17/msg031245.html

(rillig)

doc: Updated pkgtools/pkglint to 20.1.6

(rillig)

pkgtools/pkglint: update to 20.1.6

Changes since 20.1.5:

Category Makefiles must only list subdirectories in SUBDIR that actually
contain a package.  There is no need to mention the other directories.
This is the same as in the top-level Makefile, where mk/ and regress/ are
not listed.

Packages from pkgsrc-wip may have a COMMIT_MSG file with a suggested
commit message for importing the package into main pkgsrc.

(rillig)

cross/xtensa-lx106-elf-binutils: skip portability check

(rillig)

lang/php72: fix build with GNU_CONFIGURE_STRICT=yes

Most PHP modules don't know the --with-pcre-dir option.

(rillig)

regress/infra-unittests: replace double quotes with single quotes

(rillig)

biology/canu: fix build on NetBSD-8.0-x86_64

(rillig)

biology/canu: fix SUBST blocks to actually find the Java code

(rillig)

mk/subst.mk: change default value for SUBST_NOOP_OK from yes to no

This makes the SUBST blocks stricter than before, to detect outdated or
unnecessary definitions.

Filename patterns that are not affected by any of the SUBST_SED
expressions make the build fail.  It is still ok if only some of the
files from a pattern are affected and some aren't.

The latest bulk build shows that most of the build failures are fixed.
The packages that fail in that build are mostly due to other failures,
like missing C headers, wrong PLIST files, unresolved references at link
time.  There may be a few packages that still fail because of this, but
these are near the leaves of the dependency tree.

https://mail-index.netbsd.org/pkgsrc-bulk/2020/05/14/msg018919.html

(rillig)

security/zoneminder: fix build in SUBST_NOOP_OK=no mode

(rillig)

sysutils/rox: remove no-op file from SUBST_FILES

(rillig)

sysutils/salt: allow no-op SUBST block

A typical case is that PKGMANDIR is man, not share/man.  That path does
not occur in the Python files, which would then make the build fail in
SUBST_NOOP_OK=no mode.

(rillig)

sysutils/xcdroast: fix paths in SUBST_FILES

(rillig)

time/todotxt: fix build in SUBST_NOOP_OK=no mode

(rillig)

www/ap2-jk: allow no-op SUBST blocks

Since the patterns cannot be statically determined to be identity
substitutions, the explicit SUBST_NOOP_OK is needed.

(rillig)

www/moodle: remove no-op file from SUBST_FILES

(rillig)

www/swiggle: remove unnecessary SUBST block

The -O3 does not exist anymore in the Makefile, at least since 2003.
Replacing gcc with the actual compiler name is not necessary since pkgsrc
provides compiler wrappers in WRKDIR/.compiler.

(rillig)

x11/kitty: remove no-op file from SUBST_FILES

(rillig)

x11/py-wxWidgets: remove no-op file from SUBST_FILES

(rillig)

x11/xfce4-session: remove no-op SUBST block

There are no hard-coded paths in xflock4.

(rillig)

print/hplip: clean up and adjust hard-coded paths in SUBST_FILES

(rillig)

security/fail2ban: fix build with SUBST_NOOP_OK=no

(rillig)

net/netdisco: remove duplicate filename from SUBST_FILES

(rillig)

net/rabbitmq-c: remove no-op SUBST block

(rillig)

misc/ja-less: remove no-op file from SUBST_FILES

(rillig)

multimedia/fxtv: remove redundant SUBST block

It is already taken care of by patches/patch-aa.

(rillig)

multimedia/gst-plugins1-vaapi: remove no-op SUBST block

The file meson.build does not contain any @PY placeholders.

(rillig)

mail/sendmail: fix build in SUBST_NOOP_OK=no mode

(rillig)

mail/qmail-rejectutils: does not have error.h

(rillig)

mail/qmail-qfilter: does not have error.h

(rillig)

mail/qmail-conf: does not have error.h

(rillig)

mail/qmail-autoresponder: does not have error.h

(rillig)

mail/qmail-acceptutils: does not have error.h

(rillig)

mail/qconfirm: does not have error.h

(rillig)

mail/policyd-weight: properly escape dot in regular expression

Without this escaping, mk/subst.mk sees that there are no actual changes
with the default setup.  Nevertheless, mk/scripts/subst-identity.awk does
not classify the sed command as an identity transformation because there
_might_ be the text /etc/policyd-weightXconf, and the X would match the
dot.  Therefore, subst.mk aborts the build when it is in SUBST_NOOP_OK=no
mode.

(rillig)

mail/mew: clean up SUBST_FILES

(rillig)

mail/mailfront: does not have error.h

(rillig)

mail/ezmlm-idx: does not have error.h

(rillig)

mail/dbmail: clean up SUBST_FILES

(rillig)

mail/courier-imap: remove no-op files from SUBST block

(rillig)

inputmethod/ibus: allow no-op SUBST block

(rillig)

lang/openjdk7: remove redundant patch

This is already taken care of by SUBST_CLASSES+=ccspath.

(rillig)

lang/sather: allow no-op SUBST block

(rillig)

graphics/inkscape: fix build in SUBST_NOOP_OK=no mode

share/patterns/CMakeLists.txt doesn't exist anymore.

(rillig)

geography/opencpn: fix SUBST block

(rillig)

games/ztrack: fix build in SUBST_NOOP_OK=no mode

The sed command is not classified as identity substitution by
mk/scripts/subst-identity.awk since the s,,, is preceded by a regular
expression.  Since this is an edge case, it will not be implemented
there.  Otherwise the other addressing modes would also have to be
implemented.

(rillig)

mk/subst.mk: properly detect s,a,a,1 as identity substitution

Seen in games/bastet.  The 1 is not necessary though since it only
repeats the default behavior of sed.

(rillig)

emulators/xtrs: fix SUBST block for oss

Defining SUBST_FILES without a corresponding SUBST_SED makes the build
fail in SUBST_NOOP_OK=no mode.

(rillig)

devel/tcllib: remove no-op file from SUBST_FILES

(rillig)

editors/zim: remove nonexistent file from SUBST block

(rillig)

devel/bugzilla3: remove no-op file from SUBST_FILES

(rillig)

graphics/dia: remove unknown configure option

(rillig)

devel/bugzilla: allow no-op SUBST block for path to diff

(rillig)

databases/postgresql-postgis2: remove no-op SUBST block

The mentioned files don't contain any placeholder, and they are not
installed either.

(rillig)

databases/postgresql-postgis2: remove outdated PGSQL_VERSIONS_ACCEPTED

This should have been PGSQL_VERSIONS_INCOMPATIBLE from the beginning, as
the comment above that line clearly said.

(rillig)

cross/gcc-mips-current: remove no-op SUBST block

(rillig)

doc: Updated devel/acr to 1.9.2

(rillig)

devel/acr: update to 1.9.2

Changes since 1.9.0:

Fix for CHKVER

(rillig)

devel/acr: update to 1.9.0

Changes since 1.7.2:

Add CHKFLG to check for C compiler flags
Add support for LANG_V
Add CHKSYM to verify if a symbol is found in the given library
Use more markdown and update man pages

(rillig)

graphics/OpenRM: fix typo in variable name

(rillig)

removed tclticker

(rillig)

finance/tclticker: remove the package

This package got its stock quotes from web sites that were active around
2000 and that no longer exist.  It had not resolved the PREFIX
placeholder, which meant it could not even ask the web sites for the
quotes.  Therefore chances are high that this package is not used by
anyone.

(rillig)

doc: Updated pkgtools/createbuildlink to 3.18

(rillig)

pkgtools/createbuildlink: remove unused placeholder

(rillig)

pkgtools/createbuildlink: use SUBST instead of ad-hoc sed

This allows to show the actual changes by setting SUBST_SHOW_DIFF=yes,
and with SUBST_NOOP_OK=no it also demonstrates that no substitutions are
needed for substplistbasedirs.

Another thing this demonstrates is this wrong patch:

-echo >>$sedrules "s|@@PKGVERSION@@|@PKGVERSION@|g"
+echo >>$sedrules "s|@3.17@|3.17|g"

(rillig)

devel/acr: fix forgotten placeholders

(rillig)

graphics/GraphicsMagick: remove unknown configure option

(rillig)

cross/nios2: skip check for unknown configure script and portability

(rillig)

graphics/graphviz: skip check for unknown configure options

(rillig)

x11/kwindowsystem: remove no-op SUBST block

That file uses a placeholder now.

(rillig)

biology/canu: don't try to chmod system binaries

/usr/bin/chmod -R g-w /home/pbulk/build/biology/canu/work
chmod: WARNING: can't change /home/pbulk/build/biology/canu/work/.gcc/bin/cc
chmod: WARNING: can't change /home/pbulk/build/biology/canu/work/.gcc/bin/g++

Just good that serious pkgsrc builds are run as an unprivileged user,
with no chance of bricking the whole operating system.

(rillig)

x11/kdelibs4: fix no-op SUBST block

There is no path literal in kopenssl.cpp anymore.

(rillig)

print/dvipdfmx: fix path to Ghostscript

(rillig)

www/firefox60: remove no-op SUBST block

The file mozHunspell.cpp does not exist anymore.  The only file in the
directory mozilla/extensions/spellcheck that contains @prefix@ is
Makefile.in.

(rillig)

www/seamonkey: remove no-op SUBST block

The file mozHunspell.cpp does not exist anymore.  The only file in the
directory mozilla/extensions/spellcheck that contains @prefix@ is
Makefile.in.

(rillig)

print/tex-musixtex: remove redundant SUBST block

The mentioned file neither contains @LOCALBASE@ nor /usr/pkg.

(rillig)

lang/oo2c: fix build on NetBSD 8.0

(rillig)

databases/mariadb55-client: remove no-op SUBST block

None of the mentioned files actually exists.

(rillig)

devel/ruby-subversion: suppress USE_TOOLS+=perl warning

(rillig)

devel/py-subversion: suppress USE_TOOLS+=perl warning

(rillig)

cross/h8300-elf-gcc: suppress USE_TOOLS+=perl warning

(rillig)

lang/gcc3*: suppress USE_TOOLS+=perl warning

(rillig)

databases/postgresql: suppress USE_TOOLS+=perl warning

(rillig)

hs-*: add PLIST files for a few more Haskell packages

(rillig)

www/py-mod_wsgi: fix bashism in configure script

(rillig)

sysutils/xentools411: clean up SUBST_FILES

(rillig)

security/polkit: skip check for unknown configure options

(rillig)

security/cyrus-sasl: suppress USE_TOOLS+=perl warning

(rillig)

net/smokeping: fix library path

(rillig)

net/samba4: remove nonexistent file from REPLACE_PERL

(rillig)

net/powerdns: remove no-op SUBST block

(rillig)

net/netatalk22: clean up SUBST block for DVIPS

The DVIPS= appears once with quotes and once without.

(rillig)

mk/tools/bsd.tools.mk: list _USE_TOOLS in show-all-tools

This variable is used in quite a few places, which makes it interesting
enough, even though it is an implementation detail.

(rillig)

math/pari23: remove redundant SUBST block

At least on NetBSD 8.0 x86_64, this is not necessary.

(rillig)

math/boolstuff: suppress USE_TOOLS+=perl warning

(rillig)

lang/gambc: skip configure in SUBST block

The configure script does not contain these placeholders anymore.

(rillig)

lang/g95: suppress USE_TOOLS+=perl warning

(rillig)

inputmethod/skkserv: fix accidental patch for version number

The version number is supposed to be substituted in pre-configure, not in
the patch itself.  To ensure this in the future, fail fast by setting
SUBST_NOOP_OK=no.

(rillig)

devel/swig: suppress USE_TOOLS+=perl warning

(rillig)

devel/sloccount: fix hard-coded /usr/pkg

(rillig)

devel/lua-thrift: suppress USE_TOOLS+=perl warning

(rillig)

devel/libthrift_c_glib: suppress USE_TOOLS+=perl warning

(rillig)

devel/libglademm: suppress USE_TOOLS+=perl warning

(rillig)

devel/gettext: skip check for unknown configure options

(rillig)

devel/fann: fix path to fann.pc file

(rillig)

devel/ddd: suppress USE_TOOLS+=perl warning

(rillig)

devel/binutils: suppress USE_TOOLS+=perl warning

(rillig)

devel/R-Rcpp: skip nonexistent file

(rillig)

databases/unixodbc: skip check for unknown configure options

(rillig)

databases/sqlrelay: avoid USE_TOOLS+=perl warning

(rillig)

databases/ruby-sqlrelay: avoid USE_TOOLS+=perl warning

(rillig)

databases/py-sqlrelay: avoid USE_TOOLS+=perl warning

(rillig)

databases/sql-relay: suppress USE_TOOLS+=perl warning

(rillig)

databases/mysql56-client: allow no-op SUBST block

In an unprivileged bulk build on NetBSD 8.0, the WRAPPER_BINDIR was not
found in the file.  Maybe there are other systems that need it.

(rillig)

devel/git-base: git-sh-setup.sh doesn't contain "cd -P" anymore

(rillig)

graphics/netpbm: the top-level C file does not have _XOPEN_SOURCE

(rillig)

lang/llvm: remove nonexistent file from SUBST_FILES

opt-remarks has probably been replaced by remarks-shlib, which has
already been added to the list.

(rillig)

mk/configure/cmake.mk: allow no-op SUBST block

For example, graphics/openjpeg on SunOS doesn't need it.

(rillig)

www/firefox60: remove redundant SUBST block

Both packages that use this file don't have the SUBST_FILES.

(rillig)

lang/sbcl: clean up SUBST_FILES

These files use SBCL_PREFIX instead of a literal string now.

(rillig)

lang/mono: allow the SUBST block fix-prefix to not affect all files

The list of files is generated via find(1) and contains each file
individually.  Some of these files are modified, some aren't.  Those
files that aren't modified are redundant, but since they are not listed
explicitly in the package Makefile, there is no superfluous code.
Avoiding such superfluous code is the whole goal of SUBST_NOOP_OK.

(rillig)

multimedia/handbrake: remove outdated SUBST block

There is no "hb.h" anymore in the source code.

(rillig)

multimedia//gst-plugins1-gio: fix build with SUBST_NOOP_OK=no

(rillig)

multimedia/gst-plugins0.10-base: fix typo in SUBST_NOOP_OK

(rillig)

x11/gdm: remove nonexistent files from SUBST_FILES

(rillig)

sysutils/u-boot: remove no-op SUBST block

That file only contains "env python2" in the first line, which is already
handled by REPLACE_INTERPRETER.

(rillig)

databases/postgresql*: suppress USE_TOOLS+=perl warning

(rillig)

regress/infra-unittests: clean up test code, add portability

The test can now be run with bash and ksh as well.

Bash exits if a function call returns failure, the NetBSD /bin/sh
doesn't.  After reading POSIX, sections "2.9.1 Simple Commands" and "set
-e", I'm in favor of bash here.

Ksh exits in "set -eu" mode if "$@" is used but no arguments are given.

(rillig)