GCC (the GNU Compiler Collection) version 2.95 contains many improvements
and features not found in the gcc-2.7.2 and gcc-2.8.1 compilers.
C++ support is vastly improved, in particular template support is worlds
above that of gcc-2.7.2.
For more details, see <http://gcc.gnu.org/>.
WWW: http://gcc.gnu.org/
Remove lang/gcc295 which is not used by any other port and had been
unused/unusable for most of last year. Recommend lang/gcc44 instead.
Approved by: maintainer
Update CONFIGURE_ARGS for how we pass CONFIGURE_TARGET to configure script.
Specifically, newer autoconf (> 2.13) has different semantic of the
configure target. In short, one should use --build=CONFIGURE_TARGET
instead of CONFIGURE_TARGET directly. Otherwise, you will get a warning
and the old semantic may be removed in later autoconf releases.
To workaround this issue, many ports hack the CONFIGURE_TARGET variable
so that it contains the ``--build='' prefix.
To solve this issue, under the fact that some ports still have
configure script generated by the old autoconf, we use runtime detection
in the do-configure target so that the proper argument can be used.
Changes to Mk/*:
- Add runtime detection magic in bsd.port.mk
(Only the first 15 lines of the commit message are shown above )
Add symmetric CONFLICTS with lang/gcc41, lang/gcc42, and lang/gcc43 which
already conflict with this port.
On the way, adjust the IGNORE statement to comply with our standards.
PR: 117430 (partly)
Approved by: portmgr (erwin)
Unbreak on -current where _types.h got a new prerequisite (sys/cdefs.h)
by me.
Unfortunately we don't have a __FreeBSD_Version for this, so the change
isn't limited to -current, it's for >= 500039.
Famous last words: it shouldn't hurt.
Requested by: portmgr (kris)
This does compile on the Alpha--well, sometimes it does. I agree with kris'
speculation that the default CFLAGS may be causing the problem seen on the
package-building cluster. Try to work around the problem.
Does not compile on alpha. Since it's unlikely anyone is going to fix
these old compilers, drop alpha from the supported arch list. Flip over
from NOT_FOR_ARCHS to ONLY_FOR_ARCHS where used, since these ports will
not support any future architectures FreeBSD runs on.
Approved by: portmgr (self)
forced commit to note that the previous change addresses PR 51807,
submitted by Kevin Oberman, maintainer of the irrtoolset port (which
I left marked broken because it has at least one other problem)
Due to popular demant into each port which might be inserted into dependency
list by bsd.port.mk insert anti foot-shooting device, which prevents
infinite fork loop when the user defines corresponding USE_XXX in global
make.conf, command line or environment.
Similar devices should probably be inserted into ports that might be inserted
into dependency list by others bsd.foo.mk files (bsd.ruby.mk, bsd.python.mk
and so on.)
Update pkg-comment to reflect GCC's new meaning (since 2.95) and refer
to http://gcc.gnu.org instead of http://egcs.cygnus.com. Also, 2.95 is
not GNU's latest offering any longer. ;-)
Approved by: obrien (maintainer+mentor)
GCC 2.95 and later aren't called egcs (any longer), so remove three patches
that add references to egcs to the documentation.
Approved by: obrien (maintainer+mentor)
Since I cannot think of any reason one would use this port for FORTRAN or
Java compilers (the newer versions are much better); trim this port down
to just the parts we will need when GCC 3.1 becomes the -CURRENT compiler.
Use ${ECHO_CMD} instead of ${ECHO} where you mean the echo command; the ECHO
macro is set to "echo" by default, but it is set to "true" if make(1) is
invoked with the -s option while ECHO_CMD is always set to the echo command.
DEFAULT_VTABLE_THUNKS got set back to `1' from `2' in 2.95.3. Just forget
about VTABLE THUNKS and use the C++ ABI the base 2.95 G++ uses since there is
no advantage the other way.
Change the libstdc++.so version number to be more like the stock one. This
change is needed as the one I was using is too much in the range of shared
version numbers the stock libstdc++.so will have.
Do dynamic PLIST creation like the `gcc-devel' port. This is the easiest way
to deal with the differing results fix-includes gives depending on the FreeBSD
version this port is built on.
