py-gobject3: add patch to address closure / tramp on ARM64

[michaelld]

Description

py-gobject3: add patch to address closure / tramp on ARM64

NOTE: This patch is not official! It is a "best effort" workaround for the issue, and for the Python interface to GObject-Introspection only. See also the actual commit for its message.

Closes: https://trac.macports.org/ticket/62180

Type(s)

• bugfix
• enhancement
• security fix

Tested on

macOS 11.5 20G71 [Intel]
Xcode 12.3 12C33

macOS 11.6 20G165 [ARM64]
Xcode 13.0 13A233

Verification

Have you

• followed our Commit Message Guidelines?
• squashed and minimized your commits?
• checked that there aren't other open pull requests for the same change?
• referenced existing tickets on Trac with full URL?
• checked your Portfile with port lint?
• tried existing tests with sudo port test?
• tried a full install with sudo port -vst install?
• tested basic functionality of all binary files?

[macportsbot]

Notifying maintainers:
@dbevans for port py-gobject3.

[kencu]

Well. Not an easy one to sort out or test. I read through https://gitlab.gnome.org/GNOME/pygobject/-/issues/455 and I can see the troubles people are having.

It's times like this that we will feel the pain from losing Jeremy's attentions...

[michaelld]

Well. Not an easy one to sort out or test. I read through https://gitlab.gnome.org/GNOME/pygobject/-/issues/455 and I can see the troubles people are having.

It's times like this that we will feel the pain from losing Jeremy's attentions...

Indeed! (all around ...) The gist of how I read the patch is that they are replicating the underlying behavior but fixing it to actually do the right thing. It's a bit of a rabbit hole to go down (the comparison), but in my read their fix is legit -- even if just for this very specific case of GO-I and LIBFFI. I'm not sophisticated enough to work out how to patch GO-I to generically do the right thing. Hence, trying this route to see if there's any traction.

[kencu]

so this is one of those situations where we leap and then fix whatever comes next, I believe.

[michaelld]

so this is one of those situations where we leap and then fix whatever comes next, I believe.

Yup that's my take too. It's like my M1 patches for NumPy and SciPy: make it work and hope upstream fixes things soonish!

[michaelld]

Checks passes, which is always a good sign. Let's see what @dbevans says ...

[michaelld]

This port is openmaintainer, we have maintainer timeout (and then some), all CI passes, and this hack/fix is critical for GNURadio's GRC to execute on M1. Per @kencu's note: this is one of those situations where we leap and then fix whatever comes next. Thus I'm merging & we'll see what happens. 🤞

[gergomez]

This is still happening. I've found an easy way to reproduce it without gnuradio:

import gi
gi.require_version("Gtk", "3.0")
from gi.repository import Gtk

window = Gtk.Window(title="Hello World")
window.show()
window.connect("destroy", Gtk.main_quit)
Gtk.main()

using
py39-gobject3@3.38.0
py39-cairo@1.20.1
libffi@3.4.2_2
glib2@@2.64.6

Running this on macOS 12.2 in a M1 Max, I get the same behavior as reported. Backtrace attached.

If I do:

gi.require_version("Gtk", "2.0")

everything works as expected.

bt.txt

[gergomez]

In fact the bug I encountered is not because of this but because of glib2 (https://trac.macports.org/ticket/64377#ticket). Solved by removing the universal.patch and config.ed and recompiling everything from source.

[michaelld]

@gergomez thanks for that info! I'll give that a try & see if I can reduce the issue to some specific part of those files ... they are usually there for a reason, but sometimes need updating for newer releases than they were originally created

[michaelld]

Excellent! For my install I have to disable that patch, then reinstall glib2, then reinstall gtk3. Then GRC works. Nice!

[wifwucite]

Cool, sounds like you guys have found a feasible work-around. Unfortunately I am not able to follow along. Could someone post some instructions on how to get GRC up and running on an ARM64 Mac with MacPorts? I think quite a few people would appreciate this. :-) Thanks in a advance.