gcc/.cvsignore
cvsdist 662572b88c auto-import changelog data from gcc-3.4.1-8.src.rpm
Tue Aug 17 2004 Jakub Jelinek <jakub@redhat.com> 3.4.1-8
- update from gcc-3_4-branch
- PRs 13956, 16684, bootstrap/16469, c++/16175, c++/16224, c++/16273,
    c++/16489, c++/16529, c++/16618, c++/16637, c++/16698, c++/16717,
    c++/16810, c++/16853, c++/16870, c++/16904, c++/16929, c++/16964,
    libgfortran/15930, libstdc++/12658, libstdc++/14697, libstdc++/16813,
    libstdc++/16959, middle-end/16790, other/16842, preprocessor/16366,
    rtl-optimization/16490, rtl-optimization/16536, rtl-optimization/16643,
    target/16239, target/16325
- avoid making silly copies in convert_move (Jeff Law)
- make sure all files in libgcj*.jar have identical timestamps accross all
    the architectures (#128431)
- one more gcj -C fix to make sure .class files are identical between
32-bit and 64-bit targets (#128431)
- put jumptables for .gnu.linkonce.t.* sections into .gnu.linkonce.r.*
    sections instead of .rodata (#129574, PR c++/16276)
- rtti linkonce fix (H.J.Lu, PR c++/16276)
- handle filenames with embedded spaces in gcj (Elliot Lee, #129675, PR
    java/9677)
- stop using dot symbols on ppc64 (Alan Modra)
- overlap fd_aux field of ppc64 .opd entries with next .opd entry's fd_func
    if a function is not going to use r11 passed to it
- avoid building multilib libjava's - they shouldn't be needed for
    packaging and otherwise we would need all of Gtk+ installed as both
    32-bit and
64-bit development environment
Thu Aug 12 2004 Thomas Fitzsimmons <fitzsim@redhat.com>
- build GTK peers, backport libjava changes from gui-branch
- rename gjar to fastjar
2004-09-09 05:49:27 +00:00

2 lines
27 B
Plaintext