CVS log for pkgsrc/devel/gettext-lib/buildlink3.mk

Up to [cvs.NetBSD.org] / pkgsrc / devel / gettext-lib

Request diff between arbitrary revisions

Keyword substitution: kv
Default branch: MAIN

(devel/gettext-lib) Update buildlink3.mk for API now does not export shortname

gettext-lib: fix relocatable for libintl; apply some pkglint suggestions

libXpm: Add USE_EXPLICIT_LIBDEPS, set to no.

This variable controls whether we add the explicit library dependencies from
libiconv and gettext-lib, required by this package as it's a bit special in how
it handles libintl.

mk: Add OPSYS_EXPLICIT_LIBDEPS and enable on SunOS.

Most systems use GNU ld, which will happily pull in symbols required by a
program even if they are only available via implicit library dependencies.  The
SunOS linker is stricter, and if a program uses a symbol then the library that
defines that symbol must be an explicit dependency.

This mostly causes problems with libiconv and libintl, both of which Linux
bundles in its C library, so a lot of third-party software does not correctly
check for them.  Until now we've had to add many, many overrides, along with
variables such as BROKEN_GETTEXT_DETECTION which nowadays only has limited
effectiveness.

The situation appears to be getting worse, especially with software built with
meson, and so both libiconv and gettext-lib will now automatically add the
correct LDFLAGS if the OPSYS sets OPSYS_EXPLICIT_LIBDEPS=yes.

This isn't perfect.  For one it isn't really an OPSYS setting as you can try to
use GNU ld on SunOS, it just doesn't work very well.  It should also really be
done via the wrappers rather than exposing LDFLAGS, but we do not yet have an
approved patch for doing this.  However it does improve the current situation.

devel: align variable assignments

pkglint -Wall -F --only aligned --only indent -r

No manual corrections.

gettext-lib: Revert previous.

Adding gettext libraries unconditionally can cause issues in a small number of
packages which deliberately try to build multilib, where the non-default arch
ends up failing as the gettext libraries are the wrong ELF class.

This will instead be fixed directly in glib2 which appears to be the root cause
of the recent fallout.

gettext-lib: Always add -lintl on SunOS.

This effectively turns on BROKEN_GETTEXT_DETECTION=yes on SunOS and adds
-lintl whenever a package pulls in gettext-lib.  The SunOS linker is
more strict than others and will ensure that library dependencies are
explicit.  Recent changes to gettext-lib and glib made the dependency
implicit and caused a lot of fallout.

Use BUILDLINK_LDFLAGS instead of BUILDLINK_LIBS so that it actually gets
added through buildlink, not only when packages honour ${LIBS}.

reverse latest commit as suggested by obache@ by keeping BUILDLINK_FNAME_TRANSFORM
but suppressing BUILDLINK_INCDIRS for gettext-lib.
for history, please refer to the thread where this suggestion initially was made:
http://mail-index.netbsd.org/pkgsrc-changes/2012/09/01/msg077728.html

gettext-lib/buildlink3.mk adds include/gettext to INCDIRS, so don't collapse
the subdirectory gettext to avoid spewing "directory not found" warnings.

devel/gettext-lib: Change buildlink location of libintl.h

The previous change to devel/gettext-lib moved the libintl header to
include/gettext.  For platforms that don't have native libintl, this
header needs to be at include inside the buildlink directory.  This
patch makes that happen.

Contributed-by: OBATA Akio

Problem:

1) The libintl.h from gettext-lib renames the gettext functions so that
   they are prefixed with 'libintl_' (So that the functions named in
   libintl don't conflict with any that maybe in libc).

2) gettext-tools is _always_ built with gettext-lib.

3) The tools architecture may make gettext-tools a build dependence or
   someone may choose to install gettext-tools.

4) The OS has built in gettext.

5) For some reason ${PREFIX}/include is add to the compiler search path and
   libintl.h from gettext-lib is found before the system one.

The result is that all the gettext functions are renamed to have the
'libintl_' prefix but libintl is not linked with and the following happens:

${FILENAME}: undefined reference to `libintl_gettext'
${FILENAME}: undefined reference to `libintl_textdomain'
${FILENAME}: undefined reference to `libintl_bindtextdomain'
collect2: error: ld returned 1 exit status

Solution:
Hide libintl.h from gettext-lib in ${PREFIX}/include/gettext then add that
to the compiler search path when realy needed.

This should permanently fix PR's pkg/24326 pkg/36201 pkg/40153 pkg/43129
pkg/44009 and pkg/44016.

Bump PKGREVISION.

Added BUILDLINK_ABI_DEPENDS

Changes 0.18.1:
* msggrep: A '$' anchor in a regular expression now also matches the end of
  the string, even if it does not end in a newline.
* Dependencies:
  The libraries and programs are now linked with libunistring if this library
  is already installed.
* Installation options:
  The configure option --with-cvs is deprecated. The 'autopoint' program will
  now use the 'git' program by default to compress its archive. If the
  configure option --without-git is specified, 'autopoint' will not rely on
  'git', but will instead rely on a locally installed a 3 MB large archive.

Changes 0.18:
* Runtime behaviour:
  - On MacOS X and Windows systems, <libintl.h> now extends setlocale() and
    newlocale() so that their determination of the default locale considers
    the choice the user has made in the system control panels.
  - On MacOS X systems, the gettext()/dgettext()/... functions now respect the
    locale of the current thread, if a thread-specific locale has been set.
* PO file format:
  There is a new field 'Language' in the header entry.  It denotes the language
  code (plus optional country code) for the PO file.  This field can be used
  by automated tools, such as spell checkers.  It is expected to be more
  reliable than looking at the file name or at the 'Language-Team' field in
  the header entry.
  msgmerge, msgcat, msgen have a new option --lang that allows to specify
  this field.  Additionally, msgmerge fills in this new field by looking at
  the 'Language-Team' field (if the --lang option is not given).
* xgettext and PO file format:
  For messages with plural forms, programmers can inform the translators
  about the range of possible values of the numeric argument, like this:
    /* xgettext: range: 0..15 */
  This information 'range: 0..15' is stored in the PO file as a flag attached
  to the message.  Translators can produce better translations when they know
  that the numeric argument is small.
* Colorized PO files:
  msgattrib, msgcomm, msgconv, msgen, msgfilter, msggrep, msginit, msgmerge,
  msgunfmt, msguniq, xgettext now have options --color and --style, like msgcat
  has since version 0.17.
* msgmerge is up to 10 times faster when the PO and POT files are large.
  This speedup was contributed by Ralf Wildenhues.
* msgcmp has a new option -N/--no-fuzzy-matching, like msgmerge has since
  version 0.12.
* msgfilter now sets environment variables during the invocation of the
  filter, indicating the msgid and location of the messge being processed.
* xgettext now can extract plural forms from Qt 4 programs. The recommended
  xgettext command-line options for this case are:
    --qt --keyword=tr:1,1t --keyword=tr:1,2c,2t --keyword=tr:1,1,2c,3t
* xgettext --language=GCC-source now recognizes also the format strings
  used in the Fortran front-end of the GCC compiler, and marks them as
  'gfc-internal-format'.
* autopoint can now be used to update several PO directories all together.

Simply and speed up buildlink3.mk files and processing.
This changes the buildlink3.mk files to use an include guard for the
recursive include. The use of BUILDLINK_DEPTH, BUILDLINK_DEPENDS,
BUILDLINK_PACKAGES and BUILDLINK_ORDER is handled by a single new
variable BUILDLINK_TREE. Each buildlink3.mk file adds a pair of
enter/exit marker, which can be used to reconstruct the tree and
to determine first level includes. Avoiding := for large variables
(BUILDLINK_ORDER) speeds up parse time as += has linear complexity.
The include guard reduces system time by avoiding reading files over and
over again. For complex packages this reduces both %user and %sys time to
half of the former time.

Greatly simplify the logic that tries to determine whether -lintl
needs -liconv in order to satisfy linkage requirements.  This is now
patterned after the approach taken with readline and termlib.

Examples on NetBSD for a package that includes only
gettext-lib/buildlink3.mk:

PREFER_NATIVE=	yes
PREFER_PKGSRC=	# empty
# This uses the native gettext and native iconv, with:
#	BUILDLINK_LDADD.gettext == "-lintl"

PREFER_NATIVE=	yes
PREFER_PKGSRC=	iconv
# This uses the native gettext and native iconv, with:
#	BUILDLINK_LDADD.gettext == "-lintl"

PREFER_NATIVE=	yes
PREFER_PKGSRC=	gettext
# This uses the pkgsrc gettext and native iconv, with:
#	BUILDLINK_LDADD.gettext == "-lintl"

PREFER_NATIVE=	yes
PREFER_PKGSRC=	gettext iconv
# This uses the pkgsrc gettext and pkgsrc iconv, with:
#	BUILDLINK_LDADD.gettext == "-lintl -liconv"

PREFER_NATIVE=	# empty
PREFER_PKGSRC=	yes
# This uses the pkgsrc gettext and pkgsrc iconv, with:
#	BUILDLINK_LDADD.gettext == "-lintl -liconv"

PREFER_NATIVE=	iconv
PREFER_PKGSRC=	yes
# This uses the pkgsrc gettext and native iconv, with:
#	BUILDLINK_LDADD.gettext == "-lintl"

PREFER_NATIVE=	gettext
PREFER_PKGSRC=	yes
# This uses the native gettext and native iconv, with:
#	BUILDLINK_LDADD.gettext == "-lintl"

PREFER_NATIVE=	gettext iconv
PREFER_PKGSRC=	yes
# This uses the native gettext and native iconv, with:
#	BUILDLINK_LDADD.gettext == "-lintl"

Use bsd.fast.prefs.mk in some very commonly included locations.

In gettext-lib/buildlink3.mk, check whether we need to pull in the
libiconv/buildlink3.mk file based on whether or not we're using the
built-in gettext-lib or not.  This is clearer than what existed in
revision 1.33 of gettext-lib/builtin.mk and fixes the problem introduced
in revision 1.34.

Change the format of BUILDLINK_ORDER to contain depth information as well,
and add a new helper target and script, "show-buildlink3", that outputs
a listing of the buildlink3.mk files included as well as the depth at
which they are included.

For example, "make show-buildlink3" in fonts/Xft2 displays:

	zlib
	fontconfig
	    iconv
	    zlib
	    freetype2
	    expat
	freetype2
	Xrender
	    renderproto

Track information in a new variable BUILDLINK_ORDER that informs us
of the order in which buildlink3.mk files are (recursively) included
by a package Makefile.

Over 1200 files touched but no revisions bumped :)

RECOMMENDED is removed. It becomes ABI_DEPENDS.

BUILDLINK_RECOMMENDED.foo becomes BUILDLINK_ABI_DEPENDS.foo.

BUILDLINK_DEPENDS.foo becomes BUILDLINK_API_DEPENDS.foo.

BUILDLINK_DEPENDS does not change.

IGNORE_RECOMMENDED (which defaulted to "no") becomes USE_ABI_DEPENDS
which defaults to "yes".

Added to obsolete.mk checking for IGNORE_RECOMMENDED.

I did not manually go through and fix any aesthetic tab/spacing issues.

I have tested the above patch on DragonFly building and packaging
subversion and pkglint and their many dependencies.

I have also tested USE_ABI_DEPENDS=no on my NetBSD workstation (where I
have used IGNORE_RECOMMENDED for a long time). I have been an active user
of IGNORE_RECOMMENDED since it was available.

As suggested, I removed the documentation sentences suggesting bumping for
"security" issues.

As discussed on tech-pkg.

I will commit to revbump, pkglint, pkg_install, createbuildlink separately.

Note that if you use wip, it will fail!  I will commit to pkgsrc-wip
later (within day).

Update gettext to 0.14.5. Many bugfixes and improvements,
including support for relocable programs, Qt support,
separation of the PO processing functions into a separate
library and more.

Reorganise the gettext infrastructure by splitting of the
tools into devel/gettext-tools, which will be used by the tools
framework. The remaining devel/gettext package contains
gettextize and autopoint aka the infrastructure to embbed gettext
into a package.

Due to the ABI and API changes, a recursive revision bump will
follow.

Take blaim by receiving the maintainer hat.

With input from jlam@, reed@ and wiz@.

Massive cleanup of buildlink3.mk and builtin.mk files in pkgsrc.
Several changes are involved since they are all interrelated.  These
changes affect about 1000 files.

The first major change is rewriting bsd.builtin.mk as well as all of
the builtin.mk files to follow the new example in bsd.builtin.mk.
The loop to include all of the builtin.mk files needed by the package
is moved from bsd.builtin.mk and into bsd.buildlink3.mk.  bsd.builtin.mk
is now included by each of the individual builtin.mk files and provides
some common logic for all of the builtin.mk files.  Currently, this
includes the computation for whether the native or pkgsrc version of
the package is preferred.  This causes USE_BUILTIN.* to be correctly
set when one builtin.mk file includes another.

The second major change is teach the builtin.mk files to consider
files under ${LOCALBASE} to be from pkgsrc-controlled packages.  Most
of the builtin.mk files test for the presence of built-in software by
checking for the existence of certain files, e.g. <pthread.h>, and we
now assume that if that file is under ${LOCALBASE}, then it must be
from pkgsrc.  This modification is a nod toward LOCALBASE=/usr.  The
exceptions to this new check are the X11 distribution packages, which
are handled specially as noted below.

The third major change is providing builtin.mk and version.mk files
for each of the X11 distribution packages in pkgsrc.  The builtin.mk
file can detect whether the native X11 distribution is the same as
the one provided by pkgsrc, and the version.mk file computes the
version of the X11 distribution package, whether it's built-in or not.

The fourth major change is that the buildlink3.mk files for X11 packages
that install parts which are part of X11 distribution packages, e.g.
Xpm, Xcursor, etc., now use imake to query the X11 distribution for
whether the software is already provided by the X11 distribution.
This is more accurate than grepping for a symbol name in the imake
config files.  Using imake required sprinkling various builtin-imake.mk
helper files into pkgsrc directories.  These files are used as input
to imake since imake can't use stdin for that purpose.

The fifth major change is in how packages note that they use X11.
Instead of setting USE_X11, package Makefiles should now include
x11.buildlink3.mk instead.  This causes the X11 package buildlink3
and builtin logic to be executed at the correct place for buildlink3.mk
and builtin.mk files that previously set USE_X11, and fixes packages
that relied on buildlink3.mk files to implicitly note that X11 is
needed.  Package buildlink3.mk should also include x11.buildlink3.mk
when linking against the package libraries requires also linking
against the X11 libraries.  Where it was obvious, redundant inclusions
of x11.buildlink3.mk have been removed.

Include ../../converters/libiconv/buildlink3.mk in buildlink3.mk,
and let it worry about whether libiconv is built-in or not.  Remove
all references to libiconv from builtin.mk.

The logic in builtin.mk was broken and unnecessary, leading to a
build failure on at least some Linux systems (such as Debian woody
without any gettext packages installed).

Libtool fix for PR pkg/26633, and other issues.  Update libtool to 1.5.10
in the process.  (More information on tech-pkg.)

Bump PKGREVISION and BUILDLINK_DEPENDS of all packages using libtool and
installing .la files.

Bump PKGREVISION (only) of all packages depending directly on the above
via a buildlink3 include.

Garbage collect BUILDLINK_PKGBASE.<pkg> from buildlink3: it is not anymore
used since revision 1.139 of mk/buildlink3/bsd.buildlink3.mk.

Fix serious bug where BUILDLINK_PACKAGES wasn't being ordered properly
by moving the inclusion of buildlink3.mk files outside of the protected
region.  This bug would be seen by users that have set PREFER_PKGSRC
or PREFER_NATIVE to non-default values.

BUILDLINK_PACKAGES should be ordered so that for any package in the
list, that package doesn't depend on any packages to the left of it
in the list.  This ordering property is used to check for builtin
packages in the correct order.  The problem was that including a
buildlink3.mk file for <pkg> correctly ensured that <pkg> was removed
from BUILDLINK_PACKAGES and appended to the end.  However, since the
inclusion of any other buildlink3.mk files within that buildlink3.mk
was in a region that was protected against multiple inclusion, those
dependencies weren't also moved to the end of BUILDLINK_PACKAGES.

If the ${PKGBASE} of a package doesn't match the token passed to
BUILDLINK_PACKAGES, then set BUILDLINK_PKGBASE.<pkg> explicitly so that
we can map from <pkg> to BUILDLINK_PKGBASE.<pkg>.

Split out the code that deals with checking whether the software is
built-in or not into a separate builtin.mk file.  The code to deal
checking for built-in software is much simpler to deal with in pkgsrc.

The buildlink3.mk file for a package will be of the usual format
regardless of the package, which makes it simpler for packagers to
update a package.

The builtin.mk file for a package must define a single yes/no variable
USE_BUILTIN.<pkg> that is used by bsd.buildlink3.mk to decide whether
to use the built-in software or to use the pkgsrc software.

Reorder some lines so that BUILDLINK_USE_BUILTIN.<pkg> set in the
environment overrides all other settings.

Create a new variable PREFER_NATIVE that has the opposite semantics
as PREFER_PKGSRC.  Preferences are determined by the most specific
instance of the package in either PREFER_PKGSRC or PREFER_NATIVE.  If
a package is specified in neither or in both variables, then PREFER_PKGSRC
has precedence over PREFER_NATIVE.

Reorganize code so that any dependencies are checked as part of deciding
whether the software is built-in or not.  This facilitates implementing
the forthcoming PKGSRC_NATIVE variable.

Move the INCOMPAT_FOO checks to a more natural location within the
block that decides whether package FOO is built-in or not.  If the
platform is listed in IMCOMPAT_FOO, then treat FOO as being not
built-in.

If we're passing through MAKEFLAGS variables whose values may contain
spaces, use the :Q modifier instead of double-quoting the value.  This
avoids breakage when executing the just-in-time su targets.

Make PREFER_PKGSRC just yes/no or a list of packages.  This makes it
simpler to understand.

Rename BUILDLINK_PREFER_PKGSRC to PREFER_PKGSRC so that we can use its
value outside of buildlink-related files.

Support a new global variable:

BUILDLINK_PREFER_PKGSRC
	This variable determines whether or not to prefer the pkgsrc
	versions of software that is also present in the base system.

	This variable is multi-state:
		defined, or "yes"	always prefer the pkgsrc versions
		not defined, or "no"	only use the pkgsrc versions if
					needed by dependency requirements

	This can also take a list of packages for which to prefer the
	pkgsrc-installed software.  The package names may be found by
	consulting the value added to BUILDLINK_PACKAGES in the
	buildlink[23].mk files for that package.

Support BUILDLINK_DEPENDS.<pkg> being a list of values.

Add some double quotes and move a section into a protected region so we
don't keep re-evaluating it.

Use l:foo: instead of S:...: to remove a library.

Use S/+$// instead of C/\+$// to save a backslash.  Very highly
recommended by seb :)

Don't need to set BUILDLINK_PREFIX.gettext explicitly as bsd.buildlink3.mk
does it correctly for us.

Sow BUILDLINK_USE_BUILTIN.<pkg> and reap _NEED_<PKG> variables.

Re-arrange to match example buildlink3.mk file in bsd.buildlink3.mk.

Initial sprinkling of work-in-progress buildlink3.mk files for using the
buildlink3 framework.