Index | Thread | Search

From:
jaxi <jaxi@cmyksoda.cc>
Subject:
Re: NEW: games/renpy 8.5.3
To:
ports@openbsd.org
Date:
Wed, 16 Sep 2026 20:33:41 -0400

Download raw body.

Thread
Thanks for picking this up, Stuart, and for handling the re-add and 
quirks parts. Your tweaks look good to me. I'll be around to fix it if 
it breaks for at least the foreseeable future, I just can't commit to 
full MAINTAINERship.

---

Thanks,

Jaxi

On 9/15/26 9:37 AM, Stuart Henderson wrote:
> On 2026/09/14 00:29, jaxi wrote:
>> Hi @ports,
>>
>> Just pinging once again on Ren'Py.
>>
>> ---
>>
>> Thanks,
>>
>> Jaxi
>>
>> On 8/31/26 11:46 AM, Jaxi Wise wrote:
>>> Hi,
>>> Gentle ping on this one. The port I sent on 2026-08-12 still applies
>>> to a current snapshot.
>>> Re-tested today: builds from scratch, packages,
>>> installs and deinstalls cleanly, renpy the_question lint passes, and
>>> the launcher runs with audio and save/load (tested across multiple vns
>>> from different developers).
>>> Any takers for a look, or an ok?
>>> ------------------------------------------------------------------------
>>> Thanks,
>>> Jaxi
>>>
>>>
>>> On August 12, 2026 10:19:53 PM EDT, jaxi <jaxi@cmyksoda.cc> wrote:
>>>
>>>      Hi, Attached is a new port of Ren'Py 8.5.3, the visual novel
>>>      engine. https://www.renpy.org/ games/renpy was removed on
>>>      2025-10-18 ("remove py2-only ports/games/renpy (has been marked
>>>      BROKEN for over a year with no interest) and pygame_sdl2 (only
>>>      used by renpy); upstream has a newer version supporting Python 3
>>>      and newer Cython if someone wants to readd it"). This is that
>>>      re-add: upstream is Python 3 with current Cython, and pygame_sdl2
>>>      has been absorbed into the Ren'Py tree, so it is a single port
>>>      now, with no pygame_sdl2 to bring back. Notes: - One build
>>>      workaround: upstream's setup.py derives CFLAGS/LDFLAGS from
>>>      pkg-config but never CXXFLAGS, so the single C++ module misses the
>>>      SDL2 include path; the port adds the pkg-config cflags via
>>>      CPPFLAGS in do-build. Already fixed properly upstream, so this
>>>      disappears at the next release update. - One patch: a backport of
>>>      upstream's fix for Python 3.14 (where functools.partial became a
>>>      method descriptor), without which any save/persistent file
>>>      containing a datetime fails to unpickle. - The launcher, template
>>>      projects and bundled games are installed under share/renpy,
>>>      following the old port's layout. Tested on amd64/-current (Intel
>>>      HD 6000): builds from scratch, packages, installs and deinstalls
>>>      cleanly; `renpy the_question lint` passes headless from the
>>>      installed package; the launcher runs under X from the installed
>>>      package with working audio and save/load. Full commercial Ren'Py
>>>      games run start to finish on the same machine and engine version.
>>>      Known limitation: Ren'Py's optional Live2D animation support
>>>      cannot work on OpenBSD (it needs Live2D Cubism Core, a proprietary
>>>      binary-only library with no BSD build). This affects a subset of
>>>      commercial games. Noted in pkg/README. I am not able to commit to
>>>      being MAINTAINER right now, so I have left it unset. I'll be happy
>>>      if someone wants to take it. I will fix anything that comes up
>>>      during review.
>>>      ------------------------------------------------------------------------
>>>      Thank you, Jaxi Wise
>>>
> I'm not entirely thrilled about adding more things that explicitly run
> a setuptools setup.py rather than use the standard pybuild infrastructure
> after I spent a bunch of time getting rid of others (there are only a
> handful left), but I tried and it seems quite resistant to doing
> things properly. So I guess we can do that, though I won't spend much
> time trying to fix if a setuptools update breaks things.
>
> Updated with some small tweaks; remove ${MODPY_FLAVOR} and remove
> *.${PATCHORIG} instead of hardcoded *.orig.port.
>
> Done as a cvs diff with the required quirks part etc because it's a
> re-add rather than an import. Any OKs to re-add? *diff removed for length*