[Pkg-octave-devel] m68k okay, arm not
Mon, 28 Feb 2005 21:19:49 -0600
On 28 February 2005 at 09:29, John W. Eaton wrote:
| On 28-Feb-2005, Rafael Laboissiere <email@example.com> wrote:
| | The octave2.1_2.1.66-1 package was correctly built by the m68k build daemon
| | last night. The gcc/g++ version was the same as the last time it failed
| | (g++-3.3_1:3.3.5-8), so I suspect that what made the buildd succeed was the
| | change in the -O compiler flag from -O1 to -O0. This was Dirk's suggestion.
| | Thanks Dirk!
| | [To JWE:] John, you might want to make this fix upstream for the m68k-linux
| | architecture.
| I'd prefer to avoid embedding any more special compiler flags in the
| build scripts. Why is -O0 needed? Is that compiler bug?
I think that's a fair statement, and particularly for things numerical, with
some emphasis on Fortran. Though sometimes C++ fails too.
The changelogs for both R and Octave are pretty full with references to m68k.
Better to have an approximate answer to the right question than a precise
answer to the wrong question. -- John Tukey as quoted by John Chambers