On Thu, Jun 02, 2005 at 07:27:52PM +0200, Michael Schwendt wrote:
On Thu, 2 Jun 2005 07:41:54 -0400, Build System wrote:
> gcc-4.0.0-9
> -----------
> * Wed May 25 2005 Jakub Jelinek <jakub(a)redhat.com> 4.0.0-9
> - update from CVS
> - PRs c++/1016, c++/21686, libfortran/18495, libfortran/19014,
> libfortran/19016, libfortran/19106, libfortran/20074,
> libfortran/20436, libfortran/21075, libfortran/21108,
> libfortran/21354, libfortran/21376, libgcj/21637, libgcj/21703,
> libgcj/21736
> - fix overflowed constant handling (Zdenek Dvorak, #156844,
> PRs middle-end/21331, tree-opt/21293)
> - make sure slow_pthread_self is never yes for linux targets
> - fix reg-stack ICE (#158407, PR target/21716)
> - fix ICE on fortran alternate returns (#158434)
> - fix ICE on fortran functions without explicit type with implicit none
> (#158232, PR fortran/21729)
Since I don't know whether the Fedora Extras buildsys runs this compiler
already, I must assume that it does, because Amarok, which built fine some
hours ago on i386, fails with many C++ related errors now.
I cannot reproduce it with
$ rpm -q gcc
gcc-4.0.0-8
and need to wait for Rawhide mirror sync first.
This compiler is used only for FC5-ish builds ATM, FC4 uses 4.0.0-8.
But if you prove 4.0.0-9 breaks some package built while 4.0.0-8
worked, please gather details and file a bug report, so that we
can investigate it to see if it is a GCC or application bug.
Jakub