Re: OpenBSD 7.4-current port and test produces an Abort trap at callback_bad with 'callback_args string'

[Matti V. <mav...@gm...>]

On a closer inspection it seems related to the line:

simple.func_with_callback( callback_bad, 'fred' )

because replacing that with a direct call:

call_back('fred')

works as expected:

...
TEST: call C++ with Python callback_good
callback_good with 'callback_args string'
PASS callback_good returned 'good result'
TEST: call C++ with Python callback_bad
callback_bad with 'fred'
PASS callback_bad: error callback_bad error
TEST: call C++ with Python callback_raise_simple_error
callback_bad with 'callback_args string'
Abort trap (core dumped)

Is there some problem with raising exceptions with the
simple.func_with_callback?

However, I also think there might be a problem with:

TEST: call C++ with Python callback_good
callback_good with 'callback_args string'
PASS callback_good returned 'good result'

because I think it shouldn't show 'callback_args string'

On Wed, Nov 29, 2023 at 1:15 PM Matti Viljamaa <mav...@gm...>
wrote:

> I am attempting to build pycxx-7.1.8 for OpenBSD 7.4-current.
>
> I am following the Unix installation guide at:
>
> https://cxx.sourceforge.net/
>
> The final line:
>
> make -f linux.mak clean test
>
> or
>
> gmake -f linux.mak clean test
>
> in OpenBSD
>
> fails to
>
> ...
> TEST: call C++ with Python callback_bad
> callback_bad with 'callback_args string'
> Abort trap (core dumped)
> gmake: *** [linux.mak:86: test] Error 134
>
> Upon inspecting test_simple.py I've noticed that it seems to be linked to
> raising ValueError, because commenting out line 31:
>
> raise ValueError( 'callback_bad error' )
>
> in test_simple.py makes test_simple.py progress to the next test. Now I
> see:
>
> ...
> TEST: call C++ with Python callback_bad
> callback_bad with 'callback_args string'
> FAILED callback_bad None
> TEST: call C++ with Python callback_raise_simple_error
> callback_bad with 'callback_args string'
> Abort trap (core dumped)
>
> However, since the Abort traps continue showing, I speculate that this is
> about something more. Possible reasons:
>
> * pycxx-7.1.8 is not compatible with Python 3.10 (or a particular
> subversion)
> * something is different in OpenBSD compared to Linux
>
> Any ideas?
>