1 |
On Thursday 06 February 2014 09:45:10 Fabian Groffen wrote: |
2 |
> All, |
3 |
> |
4 |
> I started last FOSDEM to give the bootstrap process some desparately |
5 |
> needed love again. |
6 |
|
7 |
I'm with you, brother :) |
8 |
|
9 |
> Sofar, I've got bootstraps succeed on OpenIndiana x86 and x64. (yay) |
10 |
> |
11 |
> Currently, bootstrap is broken on: |
12 |
> - Darwin: it hangs on bash, which DEMANDS readline, despite disabling |
13 |
> it, I think the best course of action here is to actually compile |
14 |
> readline before bash, but I don't recall why we didn't do that |
15 |
> (probably because it used to break). |
16 |
|
17 |
readline in term demands ncurses, and this gets you into trouble with |
18 |
libraries I mentioned earlier: the libraries built by the host compiler may |
19 |
not be ABI-compatible with the libraries built by prefix's gcc later on. This |
20 |
means that when you later rebuild readline (or ncurses) during emerge -e |
21 |
system, you'll suddenly find yourself without a working bash. |
22 |
|
23 |
> - Solaris 10: the old gcc compiler can't compile |
24 |
> baselayout-prefix/libeinfo. I'm about to patch that piece of code |
25 |
> tonight so it will compile. |
26 |
|
27 |
There are two problems here: libeinfo uses c99 which is not available |
28 |
everywhere (and in particular not on solaris 10) (this is easily patched by |
29 |
getting rid of the stdbool.h include); and then there's the linker error, for |
30 |
which you found a workaround but which I don't understand properly. |
31 |
|
32 |
> - Fedora 17: Python 3 (whatelse) gives problems compiling modules. The |
33 |
> oddest one _decimal, which breaks on some weird funcs, but maybe those |
34 |
> are arithmetic ones from the compiler that we may or may not have due |
35 |
> to the bootstrap setup. Solution might be to block Python 3 during |
36 |
> bootstrap stages if it demands a newer compiler or something. |
37 |
|
38 |
Regarding _decimal, is this bug 490774, perhaps? That one SHOULD be fixed. |
39 |
|
40 |
For _crypt and _nis, the problem is that python incorrectly hardcodes some |
41 |
libc paths. I was planning to look into solving this properly when the RAP |
42 |
infrastructure is in place, as it probably needs somewhat different patches |
43 |
for RAP and mainline prefix. |
44 |
|
45 |
> |
46 |
> I intend on running the bootstraps again on OpenIndiana once I fixed the |
47 |
> above issues. Any help is welcome of course. Especially now the |
48 |
> readline case for Darwin just needs people to try and see if that |
49 |
> results in problems for them. |
50 |
> |
51 |
> Over the last few days I've fixed quite some issues caused by implicit |
52 |
> dependencies. One reason I think Darwin currently breaks is because it |
53 |
> doesn't come with readline (but libedit) while most other systems /do/ |
54 |
> have readline installed. Similar reason I had a problem with bison on |
55 |
> Fedora 17 because its configure demanded perl, and Fedora doesn't |
56 |
> install that by default, while most other systems have it around... |
57 |
> |
58 |
> Fabian |
59 |
|
60 |
I hit the same problem in freebsd recently, didn't look into it in detail yet. |
61 |
|
62 |
Why do we need bison in stage1, exactly? If we can get rid of that, adding |
63 |
perl to the early stage3 list should fix this one. |
64 |
|
65 |
-- Ruud |