1 |
On Wed, 23 Aug 2006, Stefan G. Weichinger wrote: |
2 |
|
3 |
> Bo Ørsted Andresen wrote: |
4 |
> > On Wednesday 23 August 2006 22:22, Michael Hennebry wrote: |
5 |
> >>> Oh, and BTW, on gentoo your optimization choices for gcc are -O, -O2 |
6 |
> >>> or nothing, because all other -O options are replaced with -O2 by |
7 |
> >>> toolchain.eclass. |
8 |
|
9 |
Perhaps I misunderstood the preceeding. |
10 |
I'd taken it to mean that recompiling any package with gentoo tools |
11 |
would result in -Os being replaced by -O2 . |
12 |
|
13 |
> >> Since the OP wanted -Os, the question remains: |
14 |
> >> How, if at all, can he get -Os ? |
15 |
> > |
16 |
> > Assuming that the OP doesn't want a broken gcc he will probably be happy |
17 |
> > with -Os for the packages that doesn't break with it... ;) |
18 |
|
19 |
What would break with -Os that wouldn't break with -O2? |
20 |
I was under the impression that both flags only allowed |
21 |
changes that didn't affect the output. |
22 |
|
23 |
> Being the OP in this case, I want to state that I didn't want "-Os for |
24 |
> all pkgs". I just decided to set "-Os" inside my CFLAGS, and I am |
25 |
> perfectly happy with any working gcc resulting from this. |
26 |
|
27 |
A statement quoted above suggested to me that |
28 |
you were unlikely to get -Os for anything. |
29 |
|
30 |
> This isn't only about control, this is also about trust: |
31 |
> I may control which settings to use for any pkg, but then I also have to |
32 |
> trust the decisions of the maintainers which choices they made for |
33 |
> individual pkgs (apart from overriding their settings, which somehow |
34 |
> questions the usage of portage IMO). |
35 |
> |
36 |
> In fact, from my point of view, I am *allowed* to trust in this. |
37 |
|
38 |
-- |
39 |
Mike hennebry@×××××××××××××××××.edu |
40 |
"it stands to reason that they weren't always called the ancients." |
41 |
-- Daniel Jackson |
42 |
|
43 |
|
44 |
-- |
45 |
gentoo-user@g.o mailing list |