Fabrizio Bertocci wrote:
Guy Harris wrote:
That's not a bug, that's a feature; in 0.99.6, the default is to enable
warnings as errors, and you have to...
If I run 'configure' with '--disable-warnings-as-errors' I don't get
that compilation error, so it must be something in the configure script
that doesn't work quite well.
...explicitly request that they *not* be treated as errors.
Hmm.. that's not what the 'configure' script says. The configure script
reports that by default
--enable-warnings-as-errors is turned off.
... and even the 'configure.in' script says:
AC_ARG_ENABLE(warnings-as-errors,
AC_HELP_STRING( [--enable-warnings-as-errors],
[Treat warnings as errors (only for gcc).
@<:@default=no@:>@]),
[
if test "x$GCC" = "xyes" -a "x$enableval" == "xyes" -a
"x$wireshark_extra_gcc_flags" != "xyes"; then
with_warnings_as_errors="yes"
AC_MSG_RESULT(yes)
else
with_warnings_as_errors="no"
AC_MSG_RESULT(no)
fi
],
with_warnings_as_errors="no"
AC_MSG_RESULT(no)
)
And the default action is to define 'with_warnings_as_errors="no"'...
so, either there is something I miss here or something is not quite right.
Also... this feature to turn all the warnings into errors, does not
apply for the packet dissectors under epan/dissectors... many of them
are full of warnings...
For the releases we (try to) turn off warnings as errors because, well,
we can't clean up all the warnings on all the platform+Glib/GTK+compiler
combinations--especially with compiler bugs.
For 0.99.6 warnings-as-errors was turned off in the top level configure
but not (by mistake) in the wiretap/configure script.