On 6/7/2020 3:04 PM, Stuart Henderson wrote:
On 2020/06/07 14:18, Klemens Nanni wrote:
On Sun, Jun 07, 2020 at 02:13:50PM +0200, Otto Moerbeek wrote:
Looking a bit closes I can see that the userland context switching
primitives are not there in
/usr/ports/pobj/boost_1_66_0/boost_1_66_0/libs/context/src so that
part is not going to fly. There might be other parts though that are
interesting and make boost-md worthwhile to build for sparc64.
Cool, thanks for looking. I'm making progress with a port using boost-md
every now and then, eventually I'll most likely run into this and can
test/report.
Meanwhile, I committed the diff such that sparc64 builds boost-md now.
I don't really like providing a library that is known to be broken.
Maybe boost-md should be split into multiple packages. Otherwise we should
disable build of the other dependent ports (kicad, icinga2, pdns_recursor,
all of which want boost-context) but it seems wrong to do it in the dependent
ports rather than the port where the problem is.
If context does not function at all then I would rather just revert the
commit . I don't want
to split this up into a bunch of sub-packages.
On another note, I see with the newest Boost release that context now
has mips64 support.
