Re: --g-fatal-warnings

[Owen Taylor <otaylor redhat com>]

Tim Janik <timj@gtk.org> writes:

> On Mon, 17 Aug 1998, Tim Janik wrote:
> 
> > hi all,
> > 
> > after Glib supporting different log levels and domains now,
> > gtk gained a few more options to specify fatal warnings.
> > glib logs g_warning() with G_LOG_LEVEL_WARNING, g_error()
> > with G_LOG_LEVEL_ERROR, and g_return*_if_fail() with
> > G_LOG_LEVEL_CRITICAL, since the latter is different from
> > a normal warning in that it actually causes abortion of
> > certain functions.
> > 
> > gtk_init() currently features:
> > --g-fatal-warnings   for abortion on g_warning in GLib
> > --gdk-fatal-warnings for abortion on g_warning in GDK
> > --gtk-fatal-warnings for abortion on g_warning in GTK+
> > --g-fatal-checks     for abortion on g_return*_if_fail in GLib
> > --gdk-fatal-checks   for abortion on g_return*_if_fail in GDK
> > --gtk-fatal-checks   for abortion on g_return*_if_fail in GTK+
> > 
> > i'd like that to reduce to either
> > 
> > --gtk-fatal-warnings for abortion on g_warning in GTK+/GDK/GLib/GModule
> > --gtk-fatal-checks   for abortion on g_return*_if_fail in GTK+/GDK/GLib/GModule
> > 
> > or
> > 
> > --gtk-fatal-checks   for abortion on g_return*_if_fail and g_warning in
> >                      GTK+/GDK/GLib/GModule
> 
> ok, i'm going to take the latter suggestion if no one objects
> (prior to the upcoming release on tuesday night).
> 
> > 
> > what are peoples opinions on this issue?
> > 
> 
> not too late to raise voices! ;)

IMO:

  --g-fatal-warnings     to abort on all g_warning and g_return*_if_fail
                         (Including application domain specific ones)

Simple, convenient, and easy to understand.

[ In any case--gtk-fatal-checks is a bad name, since if checks were fatal,
  any GTK+ progam wouldn't last long... Only failed checks are
  fatal... in this usage ]

Also, I don't see any real reason to split off g_return_*_if_fail
into a separate 'critical' warning level.

Regards,
                                        Owen