1 |
Apparently, though unproven, at 16:42 on Friday 10 June 2011, Pandu Poluan did |
2 |
opine thusly: |
3 |
|
4 |
> On Fri, Jun 10, 2011 at 05:46, walt <w41ter@×××××.com> wrote: |
5 |
> > On 06/08/2011 10:54 PM, Pandu Poluan wrote: |
6 |
> >> $ ./configure --prefix=/usr --build=x86-pc-linux-gnu |
7 |
> > |
8 |
> > Hm. That should be --build=i686-pc-linux-gnu (for 32-bit machines) |
9 |
> > |
10 |
> > My first thought would be to grep through /etc/* for x86. At this |
11 |
> > minute I don't have a second thought :( |
12 |
> |
13 |
> Amazing! You're right! |
14 |
> |
15 |
> Apparently I created a make.conf with x86-pc-linux-gnu instead of |
16 |
> i686-linux-gnu. |
17 |
> |
18 |
> I'll blame that to installing too many x86_64-pc-linux-gnu systems :-P |
19 |
> |
20 |
> ( That said, isn't it ... inconsistent that the flags are ~amd64 and |
21 |
> ~x86 while the CHOST can be x86_64 and i686 ... ) |
22 |
|
23 |
Not at all inconsistent. Stupid, but not inconsistent. |
24 |
|
25 |
AMD developed the 64 bit instructions set so it is correctly amd64. Intel |
26 |
developed the 32 bit instruction set so they can call it what they want it to |
27 |
be, which is x86. |
28 |
|
29 |
Now, as for CHOST. Well, that comes from Red Hat who are deeply involved in |
30 |
glibc and gcc and they insist that the OneAndOnlyTrueName(tm) is x86_64. |
31 |
|
32 |
Red Hat are flat out wrong of course, but that doesn't stop them punting their |
33 |
wrong idea and shoving it into the toolchain when we regular schmucks have to |
34 |
deal with it. |
35 |
|
36 |
So now you know. The stupid ones are very very consistent in their stupidity. |
37 |
|
38 |
|
39 |
-- |
40 |
alan dot mckinnon at gmail dot com |