Re: RFC: merged python exception handling patch

[Doug Evans <dje at google dot com>]

On Mon, Aug 13, 2012 at 11:02 AM, Tom Tromey <tromey@redhat.com> wrote:
>>>>>> "Doug" == Doug Evans <dje@google.com> writes:
>
> Doug> I'd move the ifndef HAVE_PYTHON into check_quit_flag, etc.
> Doug> It's odd to have them sometimes be macros and sometimes be functions.
> Doug> This could involve having, e.g., check_quit_flag calling
> Doug> py_check_quit_flag, but it's not *that* much extra code. :-)
>
> Here's an updated patch.
> Built and regtested both with and without Python.

I like my suggestion better, but ok. :-)

> +/* Return true if the quit flag has been set, false otherwise.  */
> +
> +int
> +check_quit_flag (void)
> +{
> +  int result = quit_flag;
> +
> +  quit_flag = 0;
> +  return result;
> +}

It's too bad '!' isn't a valid symbol character, if this were Scheme I
might name it "quit?!".  1/2 :-)
[there isn't an equivalent simple convention here for explicitly
denoting functions with side-effects]

I think the function comment needs to point out that quit_flag is reset.
[I realize the comment at the declaration says this.]

Also, I think you need to temporarily block SIGINT between fetching
quit_flag and resetting it.