On Thu, 2004-09-09 at 20:33, Tim Janik wrote:
> On Thu, 9 Sep 2004, Owen Taylor wrote:
>
> > On Tue, 2004-04-13 at 03:05, Tim Janik wrote:
> > > O
> > > > Which is a significant saving for such a tiny patch...
> > >
> > > looks good and makes sense to me. i'd just not call it
> > > g_return_failed_internal, that sounds as if a "return"
> > > failed. what's bad about "g_return_if_fail_internal" ?
> >
> > I want with g_return_if_fail_warning().
>
> hm, looking at this again, i think putting "_return_" into
> a function name is not a good idea, it's ok for the macros
> though, since they actually do perform a "return" for you.
> so, better name it straight:
It's an internal helper function even if it's in the public ABI...
I'm not sure I want to spend a lot of time renaming it
one way or the other.
I think that g_assertion_failed() is wrong though, because the
function isn't suitable for use in g_assert() ... for g_assert(),
you most likely want the file/line information that was removed
in this patch.
The point being that
gtk_widget_show: assertion 'widget != NULL' failed
is clear, adding gtkwidget.c:12345 adds nothing.
next_link: assertion 'l->next != NULL' failed
is close to useless.
Regards,
Owen
> +void g_assertion_failed (const char *log_domain,
> + const char *pretty_function,
> + const char *expression);
> +
> #define g_return_if_fail(expr) G_STMT_START{ \
> if G_LIKELY(expr) { } else \
> { \
> - g_log (G_LOG_DOMAIN, \
> - G_LOG_LEVEL_CRITICAL, \
> - "file %s: line %d (%s): assertion `%s' failed", \
> - __FILE__, \
> - __LINE__, \
> - __PRETTY_FUNCTION__, \
> - #expr); \
> + g_assertion_failed (G_LOG_DOMAIN, \
> + __PRETTY_FUNCTION__, \
> + #expr); \
> return; \
> };
Attachment:
signature.asc
Description: This is a digitally signed message part