This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Anyone got irix5? I have questions about your ctype implementation


 > From: Rainer Orth <ro@TechFak.Uni-Bielefeld.DE>
 > 
 > "Kaveh R. Ghazi" <ghazi@caip.rutgers.edu> writes:
 > 
 > > If you decide to bootstrap on irix5, your g++ and libstdc++ testsuite
 > > results will look bad unless you get some patches I'm working on.
 > > (Not all of the have been submitted yet.)  So let me know if you
 > > change your mind about bootstrapping it. :-)
 > 
 > In the GCC 3.0 timeframe I've been working on patches to revive the irix5
 > cofigurations and first support o32 code generation on irix6 in a separate
 > configuration, than completing my integration into the baseline irix6
 > config I had started a couple of years ago.  I don't have any irix5
 > machines left, and even the irix6 ones are so slow that bootstrapping is
 > generally a pain, though.
 > 
 > Maybe we can check what each of us is doing to avoid duplicate work.
 > 	Rainer

Sure, what versions of irix6 do you have left?  I.e. there's a big
difference between 6.5 and any 6.x prior to that.  I'm using 6.2 which
is in some ways like irix5.x and some like irix6.5.  The work I'm
doing now involves trying to fix problems above and beyond those that
appear on 6.5 since others generally get to those before I do. :-)

Listed in three categories, based on whether I have a solution and if
it has been installed, the problems I've worked on or am working on are:

1.  libstdc++-v3 8-bit ctype safety (patch installed on trunk, not 3.1.)
2.  libstdc++-v3 numeric limits fix (patch installed on trunk, not 3.1.)
3.  libstdc++-v3 ctype +1 offset bug  (patch installed on trunk, not 3.1.)

4.  Major g++ & libstdc++-v3 breakage when -ansi is used on irix6.2,
    (-ansi is used in the testsuite and most of the testsuite fails!)
    Patch developed but not yet submitted.
5.  libstdc++-v3/libsupc++ warnings from malloc et al not being
    declared potentially breaks on most 64-bit systems and those for
    which implicit int vs void* are returned in different registers.
    Patch developed but not yet submitted.

6.  Can't build libf2c on irix6.2 due to a 32/64 multilib problem
    which has been there since Nov when it was converted to libtool.
    No solution yet.
7.  libstdc++-v3 failures on irix6.2 beyond those that appear on 6.5.
    FAIL: 21_strings/ctor_copy_dtor.cc execution test
    FAIL: 21_strings/insert.cc execution test
    FAIL: 27_io/ios_base_storage.cc execution test
    No solution yet.

If one of you would like to tackle 6 or 7 that would be great!

		--Kaveh
--
Kaveh R. Ghazi			Director of Systems Architecture
ghazi@caip.rutgers.edu		Qwest Global Services


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]