Re: [PATCH 13/16] gdb: remove cmd_list_element::function::sfunc

[Simon Marchi via Gdb-patches <gdb-patches@sourceware.org>]

On 2021-07-28 3:10 p.m., Tom Tromey wrote:
>>>>>> "Simon" == Simon Marchi via Gdb-patches <gdb-patches@sourceware.org> writes:
> 
> Simon> I don't understand what the sfunc function type in
> Simon> cmd_list_element::function is for.
> 
> Thanks for cleaning this stuff up.
> My feeling is that, probably, some of this was from a planned transition
> that was never completed.  It's hard to know for sure though.

That's what I thought as well.

> Simon>  int
> Simon>  cli_user_command_p (struct cmd_list_element *cmd)
> Simon>  {
> Simon> -  return (cmd->theclass == class_user
> Simon> -	  && (cmd->func == do_simple_func || cmd->func == do_sfunc));
> Simon> +  return cmd->theclass == class_user && cmd->func == do_simple_func;
> 
> This seems to change the semantics of this function, but I wonder if it
> matters at all.  (Ideally, though, cli_user_command_p would simply go
> away.)

I'm not 100% sure either, this is all very confusing.

The ways to create commands with class_user I know of are:

 - do_define_command
 - Python and Guile, if the user passes gdb.COMMAND_USER as the class.

do_define_command passes the dummy user_defined_command function as the
command's function, which makes it have a "simple func".  Python and
Guile do not.  So in the end that check will filter out Python and Guile
commands, which I think is what we want to do.

Simon