This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: Warnings in the C++ Front-End and GCC in General


   Date: Tue, 08 Sep 1998 00:35:33 -0600
   From: Jeffrey A Law <law@cygnus.com>

[ BTW: Mark I note that you broke your own suggested rule here,
       checking in a patch before sending it here for critiqueing
       first :-) ]

     > This patch addresses a long-standing need in GCC: the ability to
     > disable individual warnings.  Until now, we have had the coarse
     > -W/-Wall/-w switches, and the -Wno- switches for controlling certain
     > categories of warnings.  However, some warnings have not been under
     > the control of individual switches.  In fact, it does not make sense
     > to add individual switches for each of the literally hundreds
     > (thousands?) of diagnostics GCC can issue!

   Well, I'm not so sure I agree this is needed or desirable.  I'll go with
   the group decision, but my gut tells me this isn't such a great idea.

I think it is not such a hot idea either.  The SGI mips compilers do
this, and not only is it cryptic (your makefiles begin to get
cluttered with cryptic numeric options scattered all over the place),
but the mips compiler people were moaning about maintaining these
numbers when I was there.

I can close my eyes when a make runs to ignore warning I don't want to
see, or if I truly do want to scan the build output then grep is
indeed my friend.  Users can do the same.  Come on, for such a special
purpose situation, how difficult is it to whip up a shell script or
whatever to grep out the warnings you don't want to hear about?

To me this classifies as something which is baggage and a maintainance
nightmare.  I don't know of anyone who really could not live without
this feature, so let's cut it off at the pass before we have to begin
maintaining it.

For the cases I've seen cited where this feature would be useful, it
seems to me to be a better indicator that our generic top level
warning options need some tweaking or that one or two ones should be
created.

Ask yourself 3 questions to understand what I am saying here:

1) How many people truly take advantage of all the configurability X11
   has at it's core?
2) How many people eat the overhead of this flexibility and never use
   %99 of it?
3) How much effort has been expended over the years maintaing it just
   because it exists in the first place?

Later,
David S. Miller
davem@dm.cobaltmicro.com


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]