🚀 go-pugleaf

In article <4ep37r$8f8@ngriffin.itc.gu.edu.au>,
	Tony Nugent <T.Nugent@sct.gu.edu.au> writes:
>davidn@sdev.blaze.net.au (David Nugent) writes:
>
>>In article <310EB9CB.3B92@flex.com.au>,
>>	David Khoury <dave@flex.com.au> writes:
>>>If you're compiling with gcc 2.7.2, you also need to be aware
>>>of a bug in the optimisation routines enabled by the -O2 flag.
>
>>This bug applies to any version of gcc in the 2.6.x and 2.7.x series
>>on ANY intel platform (not just 2.7.2).
>
>True.  If you use -fno-strength-reduce the bug can be "avoided".  It
>can also be put into your /usr/lib/gcc-lib/i486-linux/2.7.0/specs
>file...

FWIW, Slackware 3.0 comes distributed that way. New versions of 2.7.x,
however, don't.

>But someome has made a 2.7.2 version with this bug (apparently) fixed:
>
>ftp://tsx-11.mit.edu:/pub/linux/packages/GCC/gcc272-no-sr-bug.lbin.tgz

Well, that's good news - hopefully Cygnus now know about it.

BTW, can you shed more light on the problem with 2.7.2? The C compiler
seems to work fine, but C++ is not so good so far as iostreams was
concerned. iostreams is incorporated into libc since C/C++ stdio have
the same code base, but some symbols seem to not have exported, making
it unusable to the point of not being able to compile the simplest
"hello.cc".


-- 
David Nugent: davidn@blaze.net.au, davidn@unique.blaze.net.au
Unix/Linux/DOS/Win/Os2 Microsystem Technical Support [FOR HIRE!]
http://www.blaze.net.au/~davidn