1 |
On Tue, 03 Oct 2006 02:56:42 +0200 Lionel Bouton |
2 |
<lionel-dev@××××××.name> wrote: |
3 |
| Being able to tune the CFLAGS is part of one of the core principles of |
4 |
| Gentoo: let the user be in control. |
5 |
|
6 |
What? No it isn't. |
7 |
|
8 |
| <li>nss_ldap stopped working with <c>-ffast-math</c> (reported to |
9 |
| break many packages changing with the actual gcc version)</li> |
10 |
|
11 |
Uh, -ffast-math has never been and will never be a safe thing to stick |
12 |
in CFLAGS. |
13 |
|
14 |
| <li><c>-fvisibility-inlines-hidden</c> still breaks some code</li> |
15 |
|
16 |
Ditto. In C++, inlining is a *guideline* that the compiler can and will |
17 |
ignore as it sees fit. Only code specifically designed to use this flag |
18 |
should specify it, and even then it can still lead to failures because |
19 |
of the non-guaranteed nature of inlining. |
20 |
|
21 |
| <li>if you used gcc-4.0, <c>-ftree-loop-linear</c> now breaks in |
22 |
| gcc-4.1(at least with mesa)</li> |
23 |
|
24 |
Users shouldn't be sticking that in CFLAGS anyway. |
25 |
|
26 |
| <li>again for gcc-4.0 users, <c>-ftree-vectorize</c> is known to be |
27 |
| broken in gcc-4.1 (at least for x86 and ppc, amd64 users seem to be |
28 |
| safe)</li> |
29 |
|
30 |
Or that. |
31 |
|
32 |
| <li><c>-fforce-addr</c> and <c>-fweb</c> break regularly on x86 with |
33 |
| video libraries or graphic processing apps which use hand-optimised |
34 |
| ASM</li> </ul> |
35 |
|
36 |
Or that. |
37 |
|
38 |
| Users with unsupported CFLAGS (see the <uri |
39 |
| link='http://gentoo-wiki.com/CFLAGS_matrix'>CFLAGS matrix</uri> for |
40 |
| example) might want to return to safe CFLAGS (see <uri |
41 |
| link='http://gentoo-wiki.com/Safe_Cflags'>Safe CFLAGS</uri>) if recent |
42 |
| updates caused them stability problems. On the other hand, more |
43 |
| adventurous users might want to experiment with CFLAGS that didn't |
44 |
| work properly with gcc-3.4.6... As always, the user is in control. |
45 |
|
46 |
Linking to that is a very bad idea. The wiki is in control of the |
47 |
minority ricer fringe. |
48 |
|
49 |
GWN shouldn't be advocating this kind of thing at all. Here's a better |
50 |
paragraph: |
51 |
|
52 |
<p>We would like to remind you that using anything beyond -O2 |
53 |
-fomit-frame-pointer -march/-mcpu/-mtune in CFLAGS or CXXLFAGS (and |
54 |
-mieee, -mabi etc on selected archs that tell you to do this), and using |
55 |
anything at all in LDFLAGS or ASFLAGS, is pointless and will lead to a |
56 |
broken system. Your penis length is not proportional to the size of your |
57 |
CFLAGS.</p> |
58 |
|
59 |
-- |
60 |
Ciaran McCreesh |
61 |
Mail : ciaranm at ciaranm.org |
62 |
Web : http://ciaranm.org/ |
63 |
as-needed is broken : http://ciaranm.org/show_post.pl?post_id=13 |