On Sun, Sep 25, 2005 at 09:52:00PM +0200, Adeodato Simó wrote: > * Jens Peter Secher [Sun, 25 Sep 2005 13:47:20 +0200]: > > > Wouter Verhelst <[EMAIL PROTECTED]> writes: > > > > On Sat, Sep 24, 2005 at 09:55:43AM +0200, Jens Peter Secher wrote: > > >> libcrypto++5.2.1c2-6 is in needs-build state, but I am pretty sure it > > >> won't build cleanly, since it didn't on other architectures using GCC > > >> 3.4, so please remove it from the build queue. Sorry, > > > > I see that libcrypto++ previously failed to build with an internal > > > compiler error. Did you try to build it with GCC 3.4 for that reason? If > > > so, you can revert that change -- gcc 4.0 has seen a recent update which > > > should fix those ICEs. > > Are you positive about this, Wouter?
I have confirmed that five builds which previously failed now complete successfully (the three upstream PRs I filed plus flac and zlib). Does that mean they'll all succeed? Maybe not; but as this change lets 40 previously-failing tests from the test suite now succeed, I'm pretty sure many of them will now work. > libcrypto++ suffered the famous "in cp_expr_size, at > cp/cp-objcp-common.c:101" ICE that affects arm, m68k and hppa; this > is Debian Bug#323133 and upstream PR21123, which doesn't seem at all > like fixed? Okay, then I've probably been too optimistic about it in this particular case. -- The amount of time between slipping on the peel and landing on the pavement is precisely one bananosecond