1 |
Hi folks, the uclibc ebuild is currently "stable" for a 2007 version of |
2 |
uclibc. Latest ebuild in the tree is from early 2009, yet there was a |
3 |
new release in early 2010? |
4 |
|
5 |
Recently I have bumped the ebuild to use a git checkout, together with |
6 |
nptl and this is definitely their best version yet (at least on x86 and |
7 |
looks like amd64 is also getting there rapidly). I'm only speaking for |
8 |
x86, but the most recent uclibc appear to bring good support for |
9 |
hardened, apparently working nptl and best compatibility yet with glibc. |
10 |
|
11 |
To bump the in-tree ebuild to use git simply requires adjusting the |
12 |
patch version variables so that old patches are not applied and changing |
13 |
the download location to be your favourite git checkout (browse the |
14 |
tree, each commit has a link that automatically creates a tarball of |
15 |
that code version) |
16 |
|
17 |
Do we have an active maintainer for uclibc at present? What can I/we do |
18 |
to help get uclibc more up to date? Personally I think there is a case |
19 |
that this is a sufficiently niche library that it's worth pushing for at |
20 |
least a keyworded ebuild that tracks git, and I think there is even a |
21 |
case that the unmasked version should generally be the latest release |
22 |
(I'm sure stuff breaks, but on average each new release breaks less |
23 |
stuff, not more)? |
24 |
|
25 |
If we don't have a maintainer then could someone perhaps proxy |
26 |
maintain? I was previously maintaining a stack of patches on an older |
27 |
release - so far none for git, but I guess others will also be sitting |
28 |
on a pile of personal tweaks? |
29 |
|
30 |
Cheers |
31 |
|
32 |
Ed W |