1 |
I have two requests that will help to make my work incredibly easy. ;) |
2 |
|
3 |
1) start using $(get_libdir) wherever possible instead of referencing |
4 |
lib directly. get_libdir is now part of the eutils eclass and is used by |
5 |
gen_usr_ldscript in eutils. so, where previously it had: |
6 |
|
7 |
cat > ${D}/usr/lib/$1 <<"END_LDSCRIPT" |
8 |
|
9 |
it now has: |
10 |
|
11 |
cat > ${D}/usr/$(get_libdir)/$1 <<"END_LDSCRIPT" |
12 |
|
13 |
The get_libdir function will always return "lib" if using portage |
14 |
<2.0.51_pre20 or if CONF_LIBDIR isnt set, so it should be safe to use |
15 |
-everywhere-. If CONF_LIBDIR is set, it should return lib32 or lib64, |
16 |
depending on what multilib capable arch you are running on. package |
17 |
maintainers arent expected to be aware of any issues with using |
18 |
lib32/lib64 (well, unless you're an amd64 dev or helping with multilib |
19 |
on mips i guess), so feel free to slap any bugs related to lib64/lib32 |
20 |
onto the arch maintainers. ;) |
21 |
|
22 |
please mark any lib32/lib64 bugs critical so that they're sure to be |
23 |
noticed, even if they're not technically critical. *pokes amd64 devs so |
24 |
that they know this was requested and dont downgrade any of those bugs* |
25 |
|
26 |
|
27 |
2) install and use gcc 3.4 if possible. this isnt a biggie for most |
28 |
archs, i'd just like for packages to be tested with gcc 3.4 since I |
29 |
havent run into any major problems on amd64 in a while, and would like |
30 |
to know what x86 or ppc specifics fail (i expect there to be a small few |
31 |
that only fail on these archs). Note that i'm not requesting anyone |
32 |
-switch- to gcc 3.4 here, just that they test the packages they maintain |
33 |
to see if they fail, and fix them if possible. |
34 |
|
35 |
a note about c++ apps: if you dont plan on switching to gcc 3.4 |
36 |
completely, it might not be easy to test any c++ apps. 3.4 includes an |
37 |
incompatible libstdc++, and strange bugs pop up or apps fail to compile |
38 |
when one dep links to .so.5 (gcc 3.2.3->3.3.x) and another links to |
39 |
.so.6 (gcc 3.4.x). they dont -always- pop up, but they -often- do. if |
40 |
you keep gcc 3.3 around, some retarded apps still link to .so.5 even |
41 |
when compiled using 3.4 unless you move gcc 3.3's libstdc++.so out of |
42 |
the way. :/ |
43 |
gcc 3.4 is slotted, but if you switch completely, the suggestion is to |
44 |
remove 3.3.x and 3.2.x. |
45 |
|
46 |
morfic has also been kind enough to provide us with an experimental gcc |
47 |
3.4 on x86 profile for those willing to make the switch: |
48 |
profiles/default-linux/x86/2004.2/gcc34/ |
49 |
|
50 |
|
51 |
Travis Tilley <lv@g.o> |
52 |
|
53 |
-- |
54 |
gentoo-dev@g.o mailing list |