[poppler] [PATCH] Simplify the GCC compiler flags/warnings

[Albert Astals Cid]

A Diumenge, 13 de març de 2011, Pino Toscano va escriure:
> Alle domenica 13 marzo 2011, Albert Astals Cid ha scritto:
> > A Dissabte, 12 de març de 2011, Pino Toscano va escriure:
> > > Alle sabato 12 marzo 2011, Hib Eris ha scritto:
> > > > Two more remarks:
> > > > 
> > > > +if test "x$GCC" == xyes; then
> > > > +  extra_cxxflags="-Wall -Wno-write-strings -Woverloaded-virtual
> > > > \ +                  -Wnon-virtual-dtor -Wcast-align
> > > > -fno-exceptions \ +                  -fno-check-new -fno-common"
> > > > +  case "$enable_compile_warnings" in
> > > > +    yes)
> > > > +      extra_cxxflags="$extra_cxxflags \
> > > > +                      -D_XOPEN_SOURCE=600 -D_BSD_SOURCE \
> > > > +                      -W -Wno-long-long -Wundef -Wconversion
> > > > -Wpointer-arith \ +                      -Wwrite-strings
> > > > -Wformat-security \
> > > > +                      -Wmissing-format-attribute"
> > > > +      ;;
> > > > +    esac
> > > > +  CXXFLAGS="$extra_cxxflags $CXXFLAGS"
> > > > +fi
> > > > 
> > > > 
> > > > 1. In the 'yes' case, you end up with both  -Wno-write-strings
> > > > and -Wwrite-strings, it is at least not obvious which one the
> > > > compiler will pick.
> > > 
> > > The last of course.
> > 
> > I disagree with that. It will result in people sending patches to
> > move all char * to const char* which adds nothing but noise to the
> > git history.
> 
> -Write-strings is in the new "yes", which is the equivalent of the
> current "kde", so nothing changes for the default configuration.

True, but the current "kde" is something noone uses, if you name it "enable 
warnings" everyone will think it is our default "warnings" setting.

Albert

> 
> > I'd just scrap the "extra warnings" setting and by default enable the
> > ones we think make sense and that's it.
> 
> That may be another idea -- other opinions?