[patch] physname regression: Non-matching type false breakpoint
Pedro Alves
pedro@codesourcery.com
Tue Jun 7 16:58:00 GMT 2011
On Monday 06 June 2011 22:19:43, Jan Kratochvil wrote:
> On Mon, 06 Jun 2011 11:21:30 +0200, Pedro Alves wrote:
> > I have trouble parsing this sentence.
>
> I agree the original sentence by me was bogus.
>
>
> > + /* If we found a single field with that name, and we were not given
> > + a specific overload instance in COPY, accept the field, if it's
> > + really a method. */
> > + if (i1 == 1 && strchr (copy, '(') == NULL)
>
> The point was IMO not when to accept the field but rather when to reject it
> - that an additional comparison is needed. Adjusted the text.
That justifies the patch/change, but looking at the resulting code, it
looked to me a bit better to justify when can the i1 == 1 special case
be applied (the code directly under the comment).
Anyway, glass half full vs glass half empty. :-) Your version is fine
with me.
> - if (i1 == 1)
> + /* If we were given a specific overload instance in COPY defer the field
> + acceptance till the strcmp_iw verification below even if we found just
> + a single field with that name. */
I suggest a comma after "COPY", and another after "below".
--
Pedro Alves
More information about the Gdb-patches
mailing list