Bitrig uses ELF on amd64 & arm and does not have any legacy dependency on a.out.
Remove if statement to check $MACHINE_ARCH is x86_64 in-order to set $OBJECT_FMT
to ELF, preventing $OBJECT_FMT being set to a.out by mistake.
Reviewed by joerg@
This replaces calling out to an external gpg command for verification
with inline verification using the security/netpgpverify library.
Bump version to 20150901.
* Reduce unnecessary diffs in Darwin, Linux, and SunOS mods.
* Sync WARNS flags on SunOS.
* Remove obsolete modifications to bsd.sys.mk and sys.mk on Darwin. This
primarily brings in PROG_CXX support.
Tested in bulk builds on all modified platforms.
* Ensure we aren't copying overlapping memory regions. Issue found by
Matthias Ferdinand.
* Free libarchive resources after use.
* Various WARNS and build cleanups.
Tested on Linux (now builds on Ubuntu 12.04), OSX, SmartOS and NetBSD.
* Significantly reduce RAM requirements for both "install" and "update",
headline number is "pkgin install gcc47" on SmartOS/x86_64 reduces from
383MB to 14MB, as well as being significantly faster, especially on
slower hardware. See my blog post at
<http://www.perkin.org.uk/posts/reducing-ram-usage-in-pkgin.html> for
more details.
* Handle pkg_add signature verification failures.
* Support pkg_summary.xz if available.
* Don't switch to parsable output when stdout isn't a tty, which broke
various automation tools.
* Integrate patches/patch-external_progressmeter.h
Tested on NetBSD, OSX, and SmartOS.
that the illumos variants such as SmartOS & OmniOS will be labelled accordingly
rather than being lumped in under the SunOS 5.11 banner.
This was put together with the help of Joerg @ pkgsrcCon 2015
Reviewed by joerg@ bsiegert@
pkg is built on top of libpkg, a new library to interface with package
registration backends. It abstracts package management details such as
registration, remote repositories, package creation, updating, etc.
on Linux, update libnbcompat to 20150615:
- Assume a C89 environment and malloc, vprintf, memcpy, memset, memmove
to be present and working.
- Use full prototypes for isblank, snprintf, asprintf and strdup.
This avoids GCC failing to do conflicts with the builtins.
My previous commit is wrong.
From share/mk/bsd.README of NetBSD src/:
NOGCCERROR If defined, prevents passing certain ${CFLAGS} to GCC
that cause warnings to be fatal, such as:
-Werror -Wa,--fatal-warnings
(The latter being for as(1).)
In !defined(NOGCCERROR) case, -Werror should be passed to CFLAGS.
This change may break many packages.
Initial patchset to add support by rodent@
Further adjustments made based on feedback by joerg@
Tested by myself with numerous bulkbuilds thanks to Patrick Wildt @ Bitrig
Reviewed by bsiegert@ joerg@ wiz@
http://ctan.org/license/gfsl
Since this is basically the LaTeX Project Public License with one clause
removed, add it to the default acceptable licenses.
(lppl is fine with both OSI and FSF and already allowed.)
The R2pkg program, which is modeled after url2pkg, creates a new
pkgsrc package for the corresponding R package. It is intended to
help pkgsrc developers create new R packages quickly and easily.
This package was discussed in a thread beginning with the following:
http://mail-index.netbsd.org/tech-pkg/2015/02/04/msg014227.html
- Only copy $kernel if it exists, fixes warning on OSX 10.10 which moves
the kernel location to a directory we already mount, noted by J. Lewis
Muir on pkgsrc-users.
- Add NO_CHECKSUM=yes to appease pkglint.
I had used the pkglint_current tag in the past to summarize the changes
whenever I bumped the version number. In the last 7 years, this hasn't
been done, so there seems to be no need for it.
WARN: Makefile:9: Please use ${MASTER_SITE_GITHUB:=glmark2/glmark2}
instead of "https://github.com/glmark2/glmark2".
This warning is wrong because HOMEPAGE is neither a plural variable nor
does it contain a "fetch URL".
To fix this, there is a new data type FetchURL, which has one more check
than the existing data type URL, namely the MASTER_SITES abbreviation.
The existing definitions in makevars.map have been updated accordingly
to the appropriate data types.
Note: This patch is not perfect, since ideally only variables of type
"List of FetchURL" should be suggested to be replaced with MASTER_SITES,
but since currently all these variables are actually "List of FetchURL"
and not simply "FetchURL", this should be fine.
Changes:
20150124
When -n is provided to pkgin ar, rm, in, fug, packages list is shown
as one package per line so it is easier to read (feature proposed
by gdt@ in GitHub issue #41)
20150122
Fixed GitHub issue #33: clean database when repository removed
20150121
Fixed some resource leakage pointed out by coverity scan
20150119
+PRESERVE packages now correctly upgraded without pkg_add -f
20150106
Dropped percentage printout if stdout is not a tty (patch by Havard
Eidnes)
Changed pkg_add -f flag with -D as suggested by Greg Troxel
20150105
Fixed an issue with meta-packages not updating pkgdb, check for
PKG_DBDIR mtime instead
20150102
Avoid deepness > 1 when checking for dependency upgrades, should fix
PR #48561
20141229
Avoid `preserved' packages to be `autoremoved'
20141228
Fixed GH issue #1, NULL PKG_PATH (which is wrong btw) makes pkgin
segfault, strcmp() secured with safe_strcmp()
Added --with-pkginstall to configure
20141223
0.7.0
Fixed GH issue #36, automatic flag no more set when a package is
installed via pkg_add(1)
Added -p (parsable) flag for 3rd party frontends
Added statistics (patch by Youri Mouton)
Package-names-containing-a-dot fix (patch by Yamamoto Takashi)
Given that libarchive is built with large file support, pkg_install
needs it too. Otherwise libarchive and pkg_install may see different
definitions of types such as struct stat, resulting in a runtime
failure on some platforms (especially Darwin 9):
% bmake package
=> Creating binary package /usr/pkgsrc/pkgtools/pkg_install/work/.packages/pkg_install-20141227.tgz
pkg_create: cannot write to archive: tar format cannot archive this (type=00)
The reason why this was unnoticed until now is that libarchive's
"configure" has just been regenerated with Autoconf 2.69. Prior to
2.69, AC_SYS_LARGEFILE did nothing significant on Darwin so there was
no actual inconsistency between libarchive and pkg_install, which was
just coincidental.
prevents us from attempting to link the inplace libarchive against itself.
Fixes build on systems which do not include a native libarchive, and avoids
possible issues on systems which do.
Tested on SmartOS (no native libarchive) and OS X in both builtin and pkgsrc
configurations. Ok joerg@.
This update is proposed as
http://mail-index.netbsd.org/tech-pkg/2014/12/08/msg013888.html
- (revbump) convert internal to two pass, and reduce fallout
- (replace-commonbump) new command to replace 'commonbump' to
'commonbump.replaced'
- (check-cvs-diff) new command to check 'cvs diff -u' if py27,py34, ruby200
and replace it if exists
Thanks wiz@ and gdt@ for advice.
NIH_<subcommand>_OPTS variables were introduced. With their help one
can configure EVERYTHING from config file.
NIH_OPTS config variable was also introduced.
nih mark:
- new option -n was added.
nih install:
- Option -D doesn't remove temporary directory.
- New options -v and -V were added.
- NO_REQPROV_CHECK and NO_REQPROV_CHECK config variables were
removed. Use NIH_INSTALL_OPTS instead.
nih uninstall:
- does not need "nih refresh -b".
Thanks to Andrey N. Oktyabrski for the report!
Fixed:
- Finding a correct dependency now works correctly if dependency
is ambiguous.
- Unicity problem in "nih install PKGNAME" was fixed.
- Analysing file conflicts now works correctly with files
containing a space character.
- Fix in saving ~/.nih/installed_* files
- Check for version emptyness in "fake packages" was added.
- Export path to the wrapper via config file and use that to replace the
libtool call in the relink command.
- Do not drop rpath entries when processing the relink command, they are
still needed.
'..' test to print "looks fine." and exit 0. Indeed, this is what
an installed pkglint does when run in pkgtools/pkglint. But the
corresponding test does not, depending in some way on the state
left by previous tests of main().
In lieu of precise understanding, just run the adjusted '..' test
earlier so that nothing causes it to fail. This will annoy someone
later (probably me), but gets us back to green.
- PythonDependency is a basic vartype
- Wording for missing LICENSE has changed (plural -> singular)
- Unknown dependency pattern is now ERROR (used to be WARN)
- qmail package behavior has been improved a bit (surprise!)
TOOL_DEPENDS is supported by default (pkg_src_summary stage)
Support for cross-builds (not tested yet)
stage_checks:
- gen_unregistered_CONFLICTS was disabled, it is too slow.
- additional section for memory allocation failures was added.
- improvements for "missing progs", "missign header" and "library
not found sections".
distbb_chroot:
- Option -p was added and other improvements.
distbb.conf and distbb.default.conf:
- Variable OS was renamed to OPSYS
- home-made function "assign" was replaced with :=
- New config variable ULIMIT_CMD was introduced.
It defaults to "ulimit -t 1800"
upload_pkgs_all_files: fix find(1) arguments.
Thanks to Andrey Oktyabrski for the report.
Version 0.65.0
mk-configure>=0.26.0 is required for build.
pkg_grep_summary
- fix for "-t strfile" (hcreate(3) hash is not autoresizable)
- if option -r was applied, the exit status is 1 for "not match"
and 2 for unexpected failure.
Code clean-ups and improvements.
Version 0.64.0
pkg_subgraph_deps:
- significant performance improvements
pkg_cmp_summary:
- option -f was added, options -p and -P are considered deprecated
Code clean-ups in the default pkg_digger.conf
also emit packages mentioned there, if a match was found in that
Makefile.common.
Adapt revbump HOWTO slightly, but not completely, since I don't
trust the comments enough yet.
Bump version.
Leaf package.
On Linux with LIBABISUFFIX, only include lib$LIBABISUFFIX directories rather
than both, as it can cause issues with ld finding suitable libraries. Fixes
a bunch of packages.
Do it for all packages that
* mention perl, or
* have a directory name starting with p5-*, or
* depend on a package starting with p5-
like last time, for 5.18, where this didn't lead to complaints.
Let me know if you have any this time.
Strip trailing slashes on package names.
The reason for this is that the package's own buildlink3.mk file usually
matches category/package, but not category/package/ (because the only
match is "BUILDLINK_PKGSRCDIR.foo").
Bump version.
binary package to a FreeBSD binary pkg. pkgsrc binary packages keep
more metadata, and do Dewey matching on package version numbers, so
it's not possible to convert in the other direction.
The DESCR says:
This is a small utility to generate a FreeBSD pkg binary package
from a pkgsrc binary package. It is invoked simply, using a
command like:
genpkgng ssam-1.9nb1.tgz
and it will convert the pkgsrc binary package to a pkg one:
% ./genpkgng.sh perl-5.18.1nb2.tgz
Converting perl-5.18.1nb2.tgz
-rw-r--r-- 1 agc users 16597860 Apr 24 19:01 perl-5.18.1nb2.txz
%
update pkg_chk from 2.0.6 to 2.0.7, based on suggestions from PR 30893
- switch from getopt to getopts. The issue described in PR 30893 ('*' being
expanded by shell globbing) had already been worked around, but getopts
is just a nicer interface
- Extend tag parsing to support -D '*' to cause all package lines to be matched
Newer RHEL distributions have pkgconfig files split across three locations,
so add support for both share/pkgconfig and lib${LIBABISUFFIX}/pkgconfig if
LIBABISUFFIX is defined.
