Re: sparc packages?

[Valery Ushakov <uwe%stderr.spb.ru@localhost>]

On Fri, Mar 03, 2023 at 19:24:27 +0100, Romain Dolbeau wrote:

> Le ven. 3 mars 2023 ? 19:18, Valery Ushakov <uwe%stderr.spb.ru@localhost> a ecrit :
> > (diskless sparc with nfs on a macppc without wapbl, g++ that needs a
> > gig of swap does make it any easier).
> 
> When I need to build a package, I usually cheat and do it in a Qemu
> virtual machine on a fast x86-64 host. Still faster CPU-wise than a
> pair of SM61, I/O are in a different league, and you can give it 1 GiB
> of RAM to ease the pain on those 'modern' codes. It's not 100%
> reliable, but it gets there faster than the actual hardware
> nonetheless...

That probably requires some setup effort that amortizes if you need to
do it regularly, but doesn't for an one-off...  I don't mind typing
make and then coming back a day or two later :), I don't really need
that much packages on my sparc box.

The weird failure I see is cmake boostrap:

  -- Checking if compiler supports C++ make_unique
  -- Checking if compiler supports C++ make_unique - no
  ...
  CMake Error at CMakeLists.txt:101 (message):
    The C++ compiler does not support C++11 (e.g.  std::unique_ptr).


  -- Configuring incomplete, errors occurred!

but there is no errors in CMakeFiles/CMakeError.log (except for clock
skew warning, that, from rtfs, should not affect the gmake's exit
status) and manually running the test-compile commands completes
successfully.

-uwe