1 |
On 02-07-2008 08:23:54 +0200, Markus Duft wrote: |
2 |
> > |
3 |
> > While compiling Mutt on Interix, I was confronted with flock + |
4 |
> > constants |
5 |
> > not being defined. Further inspection of the header file (manpage said |
6 |
> > it should exist in fcntl.h) revealed the need for _ALL_SOURCE. Many |
7 |
> > ebuilds just include flag-o-matic to do append-flags -D_ALL_SOURCE on |
8 |
> > Interix. Not being very amused by the beauty of that solution, and |
9 |
> > looking at the insanity of the Interix headers, what's against just |
10 |
> > adding -D_ALL_SOURCE in the Interix' profile.bashrc file (with |
11 |
> > encompassing logic not to add it multiple times)? Wouldn't that be |
12 |
> > much |
13 |
> > saner (and cleaner in the ebuilds)? Are there packages known to break |
14 |
> > when _ALL_SOURCE is defined? |
15 |
> |
16 |
> I guess there where packages refusing to build with _ALL_SOURCE, but I can't think of a single one. After all it shouldn't be too much work to get those to build with _ALL_SOURCE then I think. |
17 |
|
18 |
The inverse (strip-flags -D_ALL_SOURCE) should be possible, but then |
19 |
happening quite a lot less, I think... |
20 |
|
21 |
> So from my POV, enabling should be ok. BUT: that would mean that packages will possibly change behavior, which means that eventually nothing has to stay compatible to the current builds. This is not a blocker I think, but we have to remember that it would be better to re-bootstrap most of the things, after this change, to keep everything consistent. |
22 |
|
23 |
What do you mean? I would just enable it, rebuild system (wait for +- 2 |
24 |
days) if that's ok, commit it, and then remove all append-flags |
25 |
-D_ALL_SOURCE stuff. People having a prefix shouldn't notice that, |
26 |
should they? New packages they install just should build, and I hope no |
27 |
upstream is so stupid to put #ifdef _ALL_SOURCE in their headers like |
28 |
Interix does. Quick grepping doesn't show anything like a changing |
29 |
definition or something. |
30 |
|
31 |
|
32 |
-- |
33 |
Fabian Groffen |
34 |
Gentoo on a different level |
35 |
-- |
36 |
gentoo-alt@l.g.o mailing list |