Re: [PATCH v2 1/2] [gdb/python] Introduce py_none/py_true/py_false functions

[Tom de Vries <tdevries@suse.de>]

On 5/14/26 7:46 PM, Tom Tromey wrote:
>>>>>> "Tom" == Tom de Vries <tdevries@suse.de> writes:
> 
> Tom> +/* The following three functions are refcount-safe wrappers around
> Tom> +   Py_RETURN_{NONE,TRUE,FALSE}.  Starting with python 3.12, None, True and
> Tom> +   False are immortal, and increasing and decreasing refcount for such object
> Tom> +   is a no-op, but while supporting older versions we ignore that aspect
> Tom> +   here.  */
> 
> This comment is mildly incorrect in that using gdbpy_ref<> inherently
> implies a decref.  Immortality of Python objects isn't really a concern
> at the API level.  It's an implementation detail of the Python
> interpreter.
> 

Hi Tom,

thank for the review.

I'm not 100% sure I understand what is mildly incorrect here.  In a v3 
(to be submitted soonish), I've expanded the comment a bit, so hopefully 
it'll be either correct, or it'll be easier to point out which bit is 
incorrect.

> The rest of the patch seems ok to me.  I'm not a huge fan of
> immediately-evaluated lambdas but OTOH it's probably harmless here.
> 

I've fixed that in the v3.

>      +  return gdbpy_ref<> ([] { Py_RETURN_FALSE; } ());
>      +}
>       #endif /* GDB_PYTHON_PYTHON_INTERNAL_H */
> 
> Newline between the closing brace and the #endif please.
> 

Fixed in the v3.

Thanks,
- Tom

> thanks,
> Tom