Re: riscv64 bulk build report

On 10/28/2021 4:06 AM, Jeremie Courreges-Anglas wrote:

> On Wed, Oct 27 2021, Jeremie Courreges-Anglas <jca@wxcvbn.org> wrote:
>> On Sat, Oct 23 2021, Brad Smith <brad@comstyle.com> wrote:
>>> On 10/16/2021 8:50 PM, Jeremie Courreges-Anglas wrote:
>>>
>>>> On Wed, Oct 13 2021, jca@wxcvbn.org wrote:
>>>>> Bulk build on riscv64.ports.openbsd.org
>>>>>
>>>>> Started : Mon Sep 27 06:14:45 MDT 2021
>>>>> Finished: Tue Oct 12 22:40:26 MDT 2021
>>>>> Duration: 15 Days 16 hours 26 minutes
>>>>>
>>>>> Built using OpenBSD 7.0 (GENERIC.MP) #71: Mon Sep 27 01:58:04 MDT 2021
>>>>>
>>>>> Built 8695 packages
>>>> Brad noticed the lack of devel/boost package.
>>>>
>>>> It wasn't built because of numpy missing, and numpy itself is missing
>>>> because of the lack of fortran support. Actually the lack of a lang/gcc
>>>> port for OpenBSD/riscv64.
>>>>
>>>> While I've tried to get lang/gcc/{8,11} to build and compile useful
>>>> stuff after k2k21, I don't have a working recipe yet.
>> Same situation, no lang/gcc for riscv64 yet.
>>
>>>> In the meantime
>>>> we could make numpy optional in devel/boost. I can clean up a wip diff
>>>> for this, but first, what do you folks think?
>>> Lets see what you have for a diff first.
>> I actually had two diffs, the first one was moving numpy to
>> a subpackage.
>>
>>>> (AFAIK boost_numpy the shared lib isn't used in the ports tree).
>>> Ya, I did not find anything that used the library.
>> Since nothing uses it, I propose removing the library. If someone needs
>> numpy support in boost and riscv64 still doesn't have lang/gcc, we can
>> reintroduce numpy support in a subpackage. IMO that complexity isn't worth
>> it as long as there is no consumer.
>>
>> ok?
> I'm trying another approach regarding numpy: apparently it can be built
> without fortran support. Tests seem to behave correctly so I'm dropping
> this boost diff for now.
>
> (Obviously the diff may still make sense for you if you want to stop
> providing boost_numpy).

I saw that. Thanks.

I think I'll leave things as is unless there is some other reason to do so.