 Dr. Ralf S. Engelschall
|
1c4c7f0b57
|
Welcome Year 2007!
|
19 years ago |
|
Dr. Ralf S. Engelschall
|
b6db372b5f
|
The GCC "multilib" feature is a problematic one as not all platforms where GCC thinks it can support multilib really provide all prerequisites for this. GCC usually thinks that amd64-linux platforms support "multilib" (32 and 64 bit at the same time), but with mostly all default installations this isn't the case. Similar to other platforms like ix86-solaris. So, instead of letting GCC enable multilib with its own decision, make the "multilib" feature an explicit one and actually _disable_ it by default (as within OpenPKG we don't want mix and actually cannot mix 32 and 64 bit code anyway). If one really wishes "multilib" with OpenPKG's GCC one now explicitly have to enable it with the "with_multilib" option.
|
19 years ago |
 Thomas Lotterer
|
3bc00fef9c
|
do not use multilibs (32+64bit) but stick to pure 32bit or pure 64bit thus unbreaking build on amd64-suse10 installed w/o 32bit backwards compatibility
|
19 years ago |
|
Thomas Lotterer
|
2782b0599d
|
new world order for "Packager" and "Distribution" headers
|
19 years ago |
|
Dr. Ralf S. Engelschall
|
29fa6c7e41
|
add preliminary AIX 32-bit support
|
19 years ago |
|
Dr. Ralf S. Engelschall
|
e4fc8eec6c
|
upgrade to new l_openpkg_release usage and remove special with_gcc stuff which should only exist in gccN packages
|
20 years ago |
|
Dr. Ralf S. Engelschall
|
e1a26335f2
|
fix GCC branding; don't try to patch files if there is nothing to patch; include a vendor program fix
|
20 years ago |
|
Dr. Ralf S. Engelschall
|
8240f9723d
|
upgrade from GCC 4.0.3 to 4.1.1
|
20 years ago |
Christoph Schug
|
f1edb6dbf6
|
make gcc sun4v-aware
|
20 years ago |
|
Dr. Ralf S. Engelschall
|
ad9f3a94a3
|
upgrading package: gcc 4.0.2 -> 4.0.3
|
20 years ago |
|
Dr. Ralf S. Engelschall
|
74cb9fe191
|
although share/java is not nice it is considered a shared directory for various Java JAR files (from any package) by GCJ, so remove the filesystem cleanup hack to get GCJ working out-of-the-box
|
20 years ago |
|
Dr. Ralf S. Engelschall
|
be8267b953
|
adjust copyrights for new year 2006
|
20 years ago |
|
Dr. Ralf S. Engelschall
|
d0b67ffe3f
|
upgrading package: gcc 4.0.1 -> 4.0.2
|
20 years ago |
|
Thomas Lotterer
|
c4caf98ac9
|
workaround for gcc-4.0.1/binutils problems under Solaris-x86
|
20 years ago |
|
Dr. Ralf S. Engelschall
|
38114f49af
|
crual hack to get Solaris 10/x86 (running in 32-bit mode) working on x86_64 platform by shamelessly disabling all extra optimizations and cross-platform/multi-lib settings on this platform. Everyone with a better solution is welcome
|
20 years ago |
|
Dr. Ralf S. Engelschall
|
a81cff1b68
|
try to fix building under with_fortran=yes
|
20 years ago |
|
Dr. Ralf S. Engelschall
|
c8c8cfe95a
|
add support for FreeBSD 7.0-CURRENT
|
20 years ago |
|
Dr. Ralf S. Engelschall
|
56afb587d6
|
Major Upgrade to GCC 4.0: gcc (3.4.4) becomes gcc3, gcc40 (4.0.1) becomes gcc, gcc40 is removed, gcc41 is kept
|
20 years ago |
|
Dr. Ralf S. Engelschall
|
6e41fa425b
|
cleanup amd64-linux workaround hacking: reduce 200KB of patch for a generated file by a few lines of substitutions
|
20 years ago |
|
Dr. Ralf S. Engelschall
|
c7a2dd6517
|
cleanup indentation of amd64 patch application and move solaris hack into .patch file where similar hacks already exists
|
20 years ago |
|
Thomas Lotterer
|
1954e61a2c
|
Remove "-z text" from the gcc default linker arguments under Solaris when "%{with_binutils}" == "no". This behaviour is the default in the Solaris linker. The "-z text" makes relocations in shared objects a fatal linker error. When static libraries are build, there is often (almost always) position dependent code generated. Whether the linking of shared objects against static libraries always makes sense is another question...
|
20 years ago |
Steve Weinreich
|
385cef4399
|
Added a patch for building libstdc++ under amd64-linux
|
20 years ago |
|
Dr. Ralf S. Engelschall
|
16f223121d
|
make post-adjustments more deterministically by first moving then then removing. This way the order of the partly nested directories do not matter any more.
|
21 years ago |
|
Dr. Ralf S. Engelschall
|
c5f37df130
|
use external zlib if we build gcc with Java support
|
21 years ago |
|
Dr. Ralf S. Engelschall
|
960231082d
|
upgrading package: gcc 3.4.3 -> 3.4.4
|
21 years ago |
|
Dr. Ralf S. Engelschall
|
6af1bb58d3
|
track only 3.x releases for now
|
21 years ago |
|
Dr. Ralf S. Engelschall
|
a3fb734bef
|
Adjust the copyright declaration in this (RSE's) copy of the OpenPKG source code now that the fork point for C&W was established (CVS tag OPENPKG_CW_FP), the shared copyright agreement between RSE and C&W is no longer active (as C&W no longer officially sponsors OpenPKG) and the new shared copyright from RSE can be taken over by the newly established juristic entity OpenPKG Foundation e.V. (instead of the non-juristic entity OpenPKG Project).
|
21 years ago |
|
Dr. Ralf S. Engelschall
|
48c8434c76
|
remove SSP patchset because there is no version since GCC 3.4.1 and the newest patch does no longer apply to GCC 3.4 since 3.4.2 anyway
|
21 years ago |
|
Dr. Ralf S. Engelschall
|
f8311f5a4a
|
align with vendor gcc configuration and map -pthread option to -lpthread instead of -lc_r under FreeBSD 5 (this mainly solves the problem that applications use -pthread -lpthread combination and this way get two intermixed Pthread APIs which later fail under runtime)
|
21 years ago |
|
Dr. Ralf S. Engelschall
|
1192602706
|
Happy New Year 2005
|
21 years ago |
|
Dr. Ralf S. Engelschall
|
6bacb37caf
|
seems like config.sub is wrong here, because it translates amd64 to x86_64 although gcc uses amd64 under installation time; also we require binutils 2.14
|
21 years ago |
|
Dr. Ralf S. Engelschall
|
6644c52596
|
upgrading package: gcc 3.4.2 -> 3.4.3
|
21 years ago |
|
Dr. Ralf S. Engelschall
|
4460b7e0ee
|
remove info/dir file (conflicts between packages)
|
21 years ago |
|
Dr. Ralf S. Engelschall
|
4b4eb1b884
|
fix regex
|
21 years ago |
Thorsten Hohmeier
|
f07d85bef8
|
upgrading package: gcc 3.4.1 -> 3.4.2
|
21 years ago |
|
Dr. Ralf S. Engelschall
|
8cfb3bbdbf
|
ops, this patch snippet is still required for solaris/x86 as it looks, so add it again
|
22 years ago |
|
Dr. Ralf S. Engelschall
|
08b97cf2b9
|
upgrading package: gcc 3.4.0 -> 3.4.1
|
22 years ago |
|
Dr. Ralf S. Engelschall
|
f70b627454
|
apply the same version substs as in gcc35 here too in order to reduce diff between gcc and gcc35
|
22 years ago |
|
Dr. Ralf S. Engelschall
|
8597fe4010
|
upgrade SSP patch and remove BOUNDS patch
|
22 years ago |
|
Thomas Lotterer
|
7677f05dac
|
apply knowledge from onsite gcc deployments and override threads setting for platforms with known limitations
|
22 years ago |
|
Dr. Ralf S. Engelschall
|
c6768becf6
|
modifying package: gcc-3.4.0 20040427 -> 20040512
|
22 years ago |
|
Dr. Ralf S. Engelschall
|
a88df8b37c
|
fix detection of GCC
|
22 years ago |
|
Dr. Ralf S. Engelschall
|
61ac701c61
|
The Tru64 ld(1) is using shared libraries instead of the archive libraries, because as the ld(1) manual page tells us: "When linking a shared object, ld searches all library directories for the shared object library (libx.so). If it cannot find the shared object library, it again searches all library directories for the archive library (libx.a)." OTOH there is "-oldstyle_liblookup, Cause ld to search each library directory, first for the shared object library (libx.so) and then for the archive library (libx.a)." So, normally it searches ALL directories first for a shared library and THEN it searches for an archive library. With this flag it searches PER directory, first for a shared lib in the dir and then for an archived lib in the dir, before going to the next directory. This is what we want again as the default behaviour. Submitted by: Karl Vogel <karl.vogel@seagha.com>
|
22 years ago |
Michael Schloh von Bennewitz
|
0f1027d655
|
repair for operating systems whose default shell is not bash
|
22 years ago |
|
Dr. Ralf S. Engelschall
|
d9d21cefed
|
check for gcc in advance; use this for the Tru64 hack and reuse it for the 'profiledbootstrap' option implementation
|
22 years ago |
|
Dr. Ralf S. Engelschall
|
b3b72886fd
|
add minimum Tru64 support plus optional support (via with_profile option) for building a profiled GCC
|
22 years ago |
|
Dr. Ralf S. Engelschall
|
2e0a58f3d1
|
upgrading package: gcc 3.3.3 -> 3.4.0
|
22 years ago |
|
Dr. Ralf S. Engelschall
|
6d326ae685
|
fix patch
|
22 years ago |
|
Dr. Ralf S. Engelschall
|
7bd8bc0b8a
|
upgrading package: gcc 3.3.2 -> 3.3.3
|
22 years ago |
|
Dr. Ralf S. Engelschall
|
cb3aa7ffeb
|
Simplify packaging by removing redundant (because is already the default) option "-n %{name}-%{version}" on "%setup" lines.
|
22 years ago |
