On 2/21/19 2:09 PM, Segher Boessenkool wrote:
> On Thu, Feb 21, 2019 at 09:40:08AM +0000, Andrew Haley wrote:
>> On 2/20/19 7:40 PM, Dennis Clarke wrote:
>>> finally a ray of light in this dark tunnel. Trying to get past the
>>> gmp/mpfr/mpc stage seems to be the furst hurdle. No. No that isn't
>>> true. Getting past libiberty was the first and now gmp/mpfr/mpc and
>>> who knows what horrors await past that.
>>
>> I've never had the slightest difficulty building gmp and its friends in-
>> tree, but then I've never tried to do a Canadian cross with them. I
>> would have thought that if in-tree building was broken someone would
>> have screamed by now.
>
> Not many people cross-build native compilers at all. Most people just build
> their native compilers on the native system (it is not like building GCC is
> so heavy, esp. if you don't bootstrap, which you don't do when you cross-
> build it either), or build it in a qemu or similar to get things going.
>
> Of course it *should* work, but it is no big surprise that very unusual
> configurations encounter problems no one else does ;-)
>
Something clearly is broken however I can not find where. The "broken"
may be at my chair. However no one can reply with an error in my
actions. I clearly laid out step by step everything I am doing and I do
get replies from well respected and knowledgeable folks with "hints" of
"try this" or "try that" but no one can say "there is the mistake ..
don't do that .. do this".
I think I may file a bug in the build process and work this forwards as
a build and configuration bug. Merely because it is a more formal
process and bug report process is well laid out and neatly understood.
All of this is fully laid out in :
https://gcc.gnu.org/ml/gcc-help/2019-02/msg00068.html
--
Dennis Clarke
RISC-V/SPARC/PPC/ARM/CISC
UNIX and Linux spoken
GreyBeard and suspenders optional
Attachment:
signature.asc
Description: OpenPGP digital signature
