From: Dan Gohman (gohmandj_at_[hidden])
Date: 2002-06-10 13:54:57
On Mon, Jun 10, 2002 at 09:52:35AM -0700, Ralf W. Grosse-Kunstleve wrote:
> -INLINE:none fixes this problem.
Right. This was a mistake on my part (as well as a bug in the compiler).
> Then I get heaps of messages of this kind:
> Strangely, with my toolset I do not get any of these although the only
> warning that is turned off is 1311.
> Other than that the compilations seem to be OK, but the enormous number
> of warning messages make it very difficult to spot real problems.
My setup has -fullwarn turned on by default, and it only has a few specific
warnings turned off. I'm guessing your toolset doesn't have -fullwarn
> What platform are you using for testing?
A variety of IRIX 6.5 machines with MIPSpro 7.3 .
> P.S.: Suggestions:
> sgi-abi -> irix-abi (there is non-irix sgi hardware)
> sgi-isa -> irix-isa
I'm thinking about suggesting something like this:
feature integer-model : unspecified ilp32 lp64 ; # add others as needed
to use instead of sgi-abi so that other 64-bit capable platforms can use
the same feature. For the ISA, I'll probably change sgi-isa to a free
feature (it doesn't affect link compatability), and maybe mips-isa would
be a more appropriate name.
Do any non-SGI mips platforms use `mips3' and/or `mips4'?
> The command-line switch is -64, but your feature is names n64. This
> is a bit confusing.
The name of the ABI in the documentation is `n64'. I agree it's confusing,
but calling it `n64' reflects its similarity to `n32', as opposed to `o32'.
-- Dan Gohman gohmandj_at_[hidden]
Boost-Build list run by bdawes at acm.org, david.abrahams at rcn.com, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk