Re: [update] math/qhull 2020.2

If you're ok with committing the plplot update in lockstep, I think a
commit of qhull alongside octave is probably best. There's still a few
test failures, but I think I should have something this weekend for the
octave update.

Personally, my instinct would be to keep the old qhull versions in for a
while and remove them from the port once the dust has settled, though.

On 4/15/22 08:45, Landry Breuil wrote:
> Le Fri, Apr 15, 2022 at 08:20:29AM +0200, Landry Breuil a écrit :
>> Le Thu, Apr 14, 2022 at 09:59:09PM +0200, Volker Schlecht a écrit :
>>>
>>> On 4/14/22 08:14, Landry Breuil wrote:
>>>> Le Wed, Apr 13, 2022 at 03:56:34PM +0200, Volker Schlecht a écrit :
>>>>> Attached diff updates qhull to the current stable version 2020.2
>>>>> This adds a reentrant version of the library, which is needed for recent
>>>>> versions of math/octave to make use of libqhull.
>>>>
>>>> I've tested it against my wip gdal update and it works fine.
>>>>
>>>> have you checked that existing consumers of qhull built fine against
>>>> this updated version ? eg octave/plplot/gdal ?
>>>
>>> with the attached version of the qhull diff, plplot builds fine:
>>> This one actually includes the non-reentant libqhull (which I still
>>> had as leftovers in my /usr/local/lib ...)
>>
>> mmm.. since upstream qhull deprecated the non-reentrant version, i'd
>> rather avoiding that. Apparently debian has a plplot patch to build
>> against the reentrant version, cf
>> https://salsa.debian.org/science-team/plplot/-/blob/master/debian/patches/reentrant-qhull.patch
>>
>> will test that.
>
> plplot builds with that debian patch, so afaict everything is in line to
> commit the qhull update alongside your octave update, or do it in two
> steps.
>
> Landry
>