This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: egcs-19980508 warning patches part 1/5
- To: "Kaveh R. Ghazi" <ghazi at caip dot rutgers dot edu>
- Subject: Re: egcs-19980508 warning patches part 1/5
- From: Jeffrey A Law <law at cygnus dot com>
- Date: Wed, 13 May 1998 18:58:33 -0600
- cc: egcs at cygnus dot com, wilson at cygnus dot com
- Reply-To: law at cygnus dot com
In message <199805132005.QAA23802@caip.rutgers.edu>you write:
> Regarding the ATTRIBUTE_UNUSED stuff, I'd like to continue
> marking unused stuff when, for some reason, its not possible to remove
> the unused thing outright. This helps cut down on the number of
> spurious warnings we get. Too many spurious warnings lead to people
> ignoring real warnings.
I consider the unused and signed comparison warnings the least
interesting and prime candidates for turning off by default.
Do we want to go scatter attributes all over the sources just to
eliminate warnings? Especially a warning that is extremely unlikely
to actually point us at a bug?
There's certainly value in eliminating unused parameters when we
can, but when an external interface issue forces us to have
parms which may not be used, it's unclear if taking the time and
effort to install and maintain unused attributes is ever going to
pay off.
I guess it's the basic "how far do we take the warning elimination
work". Some claim we've already gone too far, others claim we
haven't gone far enough. I'm still undecided myself, but leaning
towards we've gone "far enough" for the unused stuff.
> Correct, because for some unknown reason scan-decls.o already
> depends on scan.h. So I deemed a second dependency unnecessary. :-)
Ah. Nevermind then.
jeff