I will test this with fontforge to see if it fixes its problem.

The abort is gone, however now I get:

fontforge -lang=py -script apply_featurefile.py Lohit-Assamese.sfd Lohit-Assamese.fea
Copyright (c) 2000-2014 by George Williams. See AUTHORS for Contributors.
 License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
 with many parts BSD <http://fontforge.org/license.html>. Please read LICENSE.
 Based on sources from 09:18 UTC  2-Jul-2018-ML-D.
 Based on source from git with hash: 
Traceback (most recent call last):
  File "apply_featurefile.py", line 23, in <module>
    if len(sys.argv) < 3:
AttributeError: module 'sys' has no attribute 'argv'

D'oh, that's what I was afraid of: fontforge is relying on the fact that Py_Main() still runs all of its command line processing, even if Py_Initialize() becomes a no-op: https://github.com/python/cpython/blob/3.6/Modules/main.c#L703

While I suspect it will leak a bit of memory (but hopefully not segfault on shutdown), it's likely worth trying a variant of Victor's patch that calls pymain_init_python_main even when the runtime is already initialized.

Can you please point me to Victor's patch? I'm a bit lost in multiple places this is being discussed 😄

I just meant this PR :)

It looks like VTK is also affected by the bug this is fixing.

/usr/local/Cellar/vtk/8.1.1/bin/vtkpython -c exit()
Fatal Python error: _Py_InitializeCore: main interpreter already initialized

Current thread 0x00007fff7507e000 (most recent call first):

If I rebuild Python 3.7 with this PR, then that error goes away.

this fixes the fontforge issue

Thanks @vstinner for the PR 🌮🎉.. I'm working now to backport this PR to: 3.7.
🐍🍒⛏🤖

GH-8352 is a backport of this pull request to the 3.7 branch.

Thanks for the review @ncoghlan! I merged my PR to master and will backport it to 3.7.