Re: [RFC 0/4] Better Python safety

[Tom Tromey <tom@tromey.com>]

>>>>> "Simon" == Simon Marchi <simark@simark.ca> writes:

Simon> We have BorrowedObject:
Simon>     https://git.efficios.com/?p=babeltrace.git;a=blob;f=src/cpp-common/bt2/borrowed-object.hpp;h=3ea62072df3f0dbfbf2ad3434980c4f7f597dd39;hb=HEAD
Simon> which holds a weak reference on an object.  It can't hold nullptr,
Simon> meaning that if you receive a BorrowedObject of some kind, you know
Simon> there is an object in it.

Initially I made gdbpy_borrowed_ref reject NULL pointers, but Python
will pass NULL as the keyword argument to varargs methods.  But I think
a separate class can be done.

Simon> Finally (1), is there something specific to GDB here?  I'm wondering if we
Simon> could make it a separate project (and import it in GDB), because it's
Simon> something I would be tempted to use outside of GDB.

Yeah, it is tied into other parts of gdb to some extent.

It checks for gdb exceptions specifically and gdbpy_convert_exception
knows about QUIT and whatnot.

gdbpy_ref is just an specialization of a template in gdbsupport.

The result converters use LONGEST, ULONGEST, and gdb::unique_xmalloc_ptr
(dunno if all those are in the patches I already sent, but anyway they
are clearly needed).

The string converter uses host_charset().

I vaguely plan to eventually use gdb_python_exception to solve the
"exception denaturation" problem, which would mean integrating it more
with the existing gdb exception system.

Simon> https://nanobind.readthedocs.io/en/latest/exchanging.html#option-3-wrappers

Simon> I'm not convinced that going with this library would be easier in the
Simon> long run, but I think we should at least take a moment to check what is
Simon> out there before writing something new.

To me this looks more complicated than anything we need.
I did look at it predecessor, pybind11, a little before starting.

Tom