There is a wierd issue with base clang and llvm37 which stops libclc to
build on 10.1 i386. So disable it so we don't block 1500+ ports.
Submitted by: pkg-fallout via antoine@
- Rename the LIBDANE option DANE because that's the name of the protocol
supported by libgnutls-dane and gnutls-cli. Also clarify the option
description.
- Add an IDN option.
- libgnutls-openssl has been removed in 3.4. Some ports used this library
in their LIB_DEPENDS but no port actually required it.
- Some old API functions have been removed. Ports that used these have been
updated or patched to use the new API.
- Add a patch to print/cups to prevent overlinking of libgnutls.so.
- Bump PORTREVISION on dependent ports.
net-im/jabber: This port used the old API to give users fine grained
control over which crypto algorithms were used via a configuration file.
It's not immediately obvious how to port this to the new API so the port
always uses the defaults now.
www/hydra: Mark BROKEN. This uses more removed calls than the other ports,
is said to be alpha quality and not fully functional and has been abandoned
10 years ago.
PR: 207768
Exp-run by: antoine
Approved by: portmgr (antoine)
The "long file names" support check touches /usr/local/lib which is a
file system violation. The value is always "yes", so just pass it as
a cached value to the configure script.
Approved by: just fix it
Now that the previous changes to the cloudabi-clang package brought in
more LLVM tools, the package actually became a misnomer. Most of the
tools that it installed are not part of Clang, but of LLVM instead. My
plans of extending it to also install wrappers around man(1) and
pkgconf(1) would make the naming even worse.
The only reason that the cloudabi-clang package was created in the first
place, was because we still had some of the core CloudABI libraries part
of FreeBSD Ports. Now that these libraries are part of their own package
collection, it makes more sense to squash this package together with
cloudabi-toolchain.
The nice thing about installing a wrapper around pkgconf(1) named
${gnu_triple}-pkg-config is that Autoconf now automatically picks it up.
Running ./configure --host=${gnu_triple} is enough to make it detect the
cross compiled libraries it depends on.
Reviewed by: bapt
Differential Revision: https://reviews.freebsd.org/D5697
For CloudABI I slowly want to switch away from certain tools provided by
Binutils. For example, it would be really nice if we could switch from
GNU ld to LLD some time in the future. LLD is not ready yet. Some bug
fixes didn't manage to land into 3.8 and it doesn't support aarch64
properly. There are already some tools that we can use, such as nm,
objdump, size, ar and ranlib.
Introduce a BUREMOVE variable that can be set in Binutils slave ports to
remove specific utilities that are installed by default. There doesn't
seem to be any other elegant way to solve this. Set a bunch of utilities
by default and then extend cloudabi-clang to set up symlinks to the LLVM
versions of the tools.
Reviewed by: bapt
Differential Revision: https://reviews.freebsd.org/D5684
There are no more ports dependent on gcc-aux anymore, and there's no
known reason to use this compiler over the newer gcc5-aux and gcc6-aux
compilers. Ideally, there should only be two compilers featured in the
Ada framework: The default and the upcoming one.
The gnatdroid cross-compilers (*BSD => Android) were the last two users
of gcc-aux (gcc 4.9 Ada compiler). They are now based on gcc6-aux.
The Android NDK release 11 just came out with support for Android API 23
(Marshmellow) so now gnatdroid supports it. I removed API 8 and 9
(Froyo and Gingerbread) from gnatdroid-armv7. Froyo currently is installed
on less than 0.1% of Android devices, and Gingerbread is dropping under 2%
soon. Many sysroot distfiles were re-rolled to gain required headers, so
a distfiles subdirectory was added as well.
When pypy isn't installed then the port needs to use python to translate
pypy, however all options are not specified since two of the three (the
pypy related options) are not valid. Overcome this by a bit of re-ordering
and abusing OPTIONS_SLAVE in the later case.
Reported by: Andrzej Tobola <ato@iem.pw.edu.pl>
Change names of gnatdroid-support dejagnu exp files so that gcc6-aux
can co-exist with gcc-aux. The gnatdroid compilers will be rebased
on gcc6-aux in the future. Technically the resulting package changes
and thus requires a bump, but nothing uses these two new files yet,
so I'm not bumping it.
Changes:
- Drop local mirror
- Drop DIST_SUBDIR (distfile name no longer a generic "release-x.y.z.tar.bz2".
- Add test dependencies (all sub-ports, a la gdbm, sqlite and tkinter)
- Use gmake instead of using sed to transform the Makefile
- Use options helper variables
- Rename regression-test -> do-test
ChangeLog (see http://morepypy.blogspot.co.za/2016/03/pypy-50-released.html):
- Faster and leaner: improved JIT warmed (up to 30% faster) and improved memory
consumption (dropped by about 30%).
- C-API Upgrade: cpyext is much faster and lxml now passes all tests when
using cython compiled components.
- CFFI: ships with cffi-1.5.2.
Fix distinfo for the offending ports.
lang/yorick's tag was moved, and the added patch was no longer needed.
PR: 207644
Submitted by: mat
Exp-run by by: antoine
Sponsored by: Absolight
Differential Revision: https://reviews.freebsd.org/D4268
Changes:
- The function fltCmp in flt_rtl.c has been improved to work correct
in the case that isnan returns neither 0 nor 1 (Many thanks go to
Arnel L. for reporting this problem).
- The function determineGetaddrlimit in chkccomp.c has been improved
to include <sys/types.h> before <sys/resource.h>. FreeBSD needs
that to work correctly (Many thanks go to Pietro Cerutti for
providing a fix and explaining the background).
- The function getOsCwd has been added to cmd_rtl.c. With this
function it is possible to read (e.g. under Linux) a current
working directory that is longer than PATH_MAX.
- Documentation comments have been improved in encoding.s7i.
- The program chkccomp.c has been improved to write definitions of
USE_DIRENT, USE_DIRWIN, OS_WIDE_DIR_INCLUDE_DIR_H, os_DIR,
os_dirent_struct, os_opendir, os_readdir, os_closedir,
INCLUDE_SYS_UTIME, os_utimbuf_struct, USE_ALTERNATE_UTIME,
os_utime_orig, os_utime, os_chdir, OS_GETCWD_MAX_BUFFER_SIZE,
os_getcwd, os_mkdir, os_rmdir, OS_CHMOD_INCLUDE_IO_H, os_chmod,
os_remove, os_rename, os_system, os_fopen, os_popen, os_pclose,
DEFINE_WGETENV, os_getenv
- The functions keys_helem and values_helem in hshlib.c and
hsh_rtl.c have been improved to use a loop instead of a tail
recursion. For compilers that don't recognize the tail recursion
this results in a speed improvement (Gcc recognizes tail
recursions so there is no speed improvement).
- The functions free_helem and free_hash in hsh_rtl.c have been
improved to stop searching for helems as soon as the last helem
has been freed.
- The functions helem_data_to_list and helem_key_to_list have been
improved to use a loop instead of a tail recursion (see above).
There was a GSoC 2015 from Colin Lord to bring to the base system a recent
version of Ficl. pfg@ found many bugs in ficl4.
Among them:
- Fix ficlInstructionPick bug. Pick instructions from 0, not 1.
Credit to Toomas Soome from the Illumos project.
- Fix some math bugs.
Credit to Toomas Soome from the Illumos project.
- Fix bug in Ficl stack preventing the stack from growing.
Obtained from: http://sourceforge.net/p/ficl/mailman/message/26634755/
- Change rand and srand calls to random and srandom
Fix those in the port. While here, unset MAKE_JOBS_UNSAFE as the port builds
fine without it, and get rid of the DATA option that has no effect.
PR: 207041
Submitted by: pfg
Approved by: Pavel Volkov <pavelivolkov@gmail.com> (maintainer)
