| 1 |
On Fri, 30 May 2008 15:07:43 -0700 |
| 2 |
Donnie Berkholz <dberkholz@g.o> wrote: |
| 3 |
> On 22:53 Fri 30 May , Ciaran McCreesh wrote: |
| 4 |
> > On Sat, 31 May 2008 00:47:44 +0300 |
| 5 |
> > Mart Raudsepp <leio@g.o> wrote: |
| 6 |
> > > The story that matters here is, that a C++ corner case that does |
| 7 |
> > > not work on 0.01% of packages with --as-needed and breaks on |
| 8 |
> > > non-ELF platforms, should not cause good things for our users to |
| 9 |
> > > be shot down. |
| 10 |
> > |
| 11 |
> > You could say the same thing for -ffast-math... |
| 12 |
> |
| 13 |
> When there's a feature that only breaks one package that we know of, |
| 14 |
> wouldn't it make more sense to enable it globally and add an |
| 15 |
> exception than to do it the other way around? |
| 16 |
|
| 17 |
Both -ffast-math and --as-needed make the compiler / linker violate |
| 18 |
various standards in ways that can't be used safely unless a package |
| 19 |
has been explicitly designed to work with it. For packages that have |
| 20 |
been explicitly designed to work with either, upstream can add the |
| 21 |
options to the build system themselves. For packages that haven't, it's |
| 22 |
not Gentoo's place to try to guess whether upstream has designed their |
| 23 |
software with ricer flags in mind, and whether if it works by fluke |
| 24 |
now it'll still work in the next version. |
| 25 |
|
| 26 |
> I see that a number of packages in the tree explicitly filter |
| 27 |
> -ffast-math. |
| 28 |
|
| 29 |
That's mostly from the bad old days when users were encouraged to use |
| 30 |
silly CFLAGS... |
| 31 |
|
| 32 |
-- |
| 33 |
Ciaran McCreesh |
Archives