1 |
On Tue, 2004-06-22 at 17:27, Peter S. Mazinger wrote: |
2 |
> On 22 Jun 2004, Ned Ludd wrote: |
3 |
> |
4 |
> > On Tue, 2004-06-22 at 13:47, Peter S. Mazinger wrote: |
5 |
> > > On 22 Jun 2004, Ned Ludd wrote: |
6 |
> > > |
7 |
> > > > > - uclibc should be a cvs version and not 0.9.26, else you'll have binary |
8 |
> > > > > compatibility issues |
9 |
> > > > Sure we can do that. Does 0.9.26.YYYYMMDD work fine for this? |
10 |
> > > |
11 |
> > > do you mean, we should call it like this? |
12 |
> > |
13 |
> > I was asking for your input on if we should call it that way if it's |
14 |
> > going to break bin compat problems. |
15 |
> |
16 |
> Well, does some else use uclibc in gentoo land (for building smaller |
17 |
> binaries for ex?). I think not, and than the name of our first "baby" is |
18 |
> not so important ;) |
19 |
> with your naming it would be more conformant to glibc... |
20 |
> |
21 |
|
22 |
|
23 |
> apropo, what about moving uclibc to sys-libs? |
24 |
That's been in the back of my minds for many months. |
25 |
Short answer is. as soon as it can be used as a sys-lib then I/we will |
26 |
move it. |
27 |
|
28 |
|
29 |
|
30 |
> |
31 |
> > > |
32 |
> > > > > 5. gcc-3.3.4 won't be supported (at least by me) until it gets a branch |
33 |
> > > > > update including the pie infrastructure (only 3.4.0 has it), or someone |
34 |
> > > > > backports/forwardports it. |
35 |
> > > > ok we can add the patch to 3.3.3-r6 |
36 |
> > > > Anything else pending for that gcc? Like the below #6 |
37 |
> > > |
38 |
> > > the stuff already added to 3.3.4 (you have sent an uclibc.patch) |
39 |
> > |
40 |
> > |
41 |
> > > would you accept the uclibc patches like I have done it for gcc-3.4.0? |
42 |
> > > only stuff w/o autoconf results, running autoconf from ebuild? It is also |
43 |
> > > not critical (as mentioned in 3.4.0 ebuild) for use build, because we do |
44 |
> > > not build at that time libstdc++ , and only this is needing autoconf. |
45 |
> > |
46 |
> > I've not looked at any of the 3.4.x stuff yet. Might try to do later.. |
47 |
> |
48 |
> look into the ebuild, where I first copy files from locale/gnu to |
49 |
> locale/uclibc and os/gnu-linux to os/uclibc and patch them later |
50 |
> this allows me to track the changes in mainline and have smaller patches |
51 |
> |
52 |
> What do you think of starting uclibc only w/ the latest ~arch packages |
53 |
> (it's a trouble, if we patch the stable ebuilds, all the newer have to be |
54 |
> patched too, else the newer ones won't have the uclibc support in), it |
55 |
> already happened for openssl that the uclibc support was not added to the |
56 |
> newer ebuild. |
57 |
> |
58 |
> The only problem I see w/ ~arch, that it can't be defined in the profile |
59 |
> in ACCEPT_KEYWORDS |
60 |
It's a pretty standard thing for developers to run ~x86 (I do..) I'd |
61 |
rather start with ~arch for * then move it back to stable or push for |
62 |
said pkg to go stable after it's been tested etc.. |
63 |
|
64 |
|
65 |
> |
66 |
> Peter |
67 |
-- |
68 |
Ned Ludd <solar@g.o> |
69 |
Gentoo (hardened,security,infrastructure,embedded,toolchain) Developer |